Home page

Main topics
News Headlines

Frequently asked Questions
For Policy Makers

Research evidence
Misleading claims
Helmet laws

Search Engine

New Zealand
Other countries

Full index

Policy statement

Head Injuries and Helmet Laws in Australia and New Zealand

Robinson DL

Last revised April 2012

Victoria was the first Australian State to introduce bicycle helmet laws, on 1 July 1990.  Over the next few  years, all other States passed similar legislation, because of threats by the Federal Government to reduce  road funding if States failed to comply with a 10-point road safety program including bicycle helmet laws.   New Zealand (NZ) also introduced a bicycle helmet law in January 1994. (Carr, Dyte and Cameron, 1995, Robinson, 2001, Hendrie, Legge, Rosman and Kirov, 1999, Marshall and White, 1994, Robinson, 1996)

Did the laws work?

The effect of the laws can be determined  by comparing injury rates before and after  the laws were introduced.


Fig1 shows numbers of  cyclists admitted to hospital with and without  head injuries before and after the helmet  law.  Both head and non-head injuries fell  substantially.  As shown later (section 3),  surveys revealed lots of people were put off  cycling.  Fewer cyclists should mean fewer  injuries.  But if helmets were effective, head  injuries should have fallen by more  than non-head injuries.  Did they?   Which line represents head injuries  and which non-head injuries?

New Zealand

Fig 2 compares  head injury (%HI) rates of primary  school children and adults admitted  to hospital.  Most primary school children were already wearing helmets,  so the law should have had little  effect on their %HI. However, adult  helmet wearing (%HW) increased  from 43% to 92%. If compulsory  wearing is beneficial, there should  have been a large reduction in %HI  of adults compared to primary school  children.  Fig 2 shows there wasn't;  both followed similar declining trends.   What other explanation is there, except that  helmets are less effective at preventing head  injury than most people think?

Western Australia

Fig 3 shows percentages of hospital admissions involving head injury for all road users.  Helmet wearing at the start of the data series was virtually nil, increasing to about 39% of cyclists just before the law was enacted on 1 Jan 1992, when helmet wearing increased to over 80%.  The most dominant feature in Fig 3 is the declining trend in %HI common to all road users.  Researchers in Victoria found a similar trend, but mistakenly concluded helmets were remarkably effective.  They didn’t bother to check that the same trend was evident for pedestrians, so had nothing to do with helmets!  If helmet laws are effective, it should be obvious from the WA data when %HW increased from 39% to more than 80%.   Can you tell which year it was?  (a later section has more details.)

South Australia

(SA, Fig 4) shows declining trends in hospital  admissions for concussion, but not other head/face injuries, and  again no obvious effect of a law that increased helmet wearing from  40-90%.  The decrease in concussions was noted and explained:   

"it is understood that, since helmet wearing became compulsory, the  procedure for patients with a short episode of concussion has  changed in that such patients are not now admitted routinely." (Marshall and White, 1994)

New South Wales

(NSW, Fig 5) introduced a helmet law for children  6 months after the law for adults.  Numbers of cyclists admitted to hospital  for head and other injuries were provided by NSW Health (Robinson, 1996).   As in other states, the dominant feature is a declining trend in  %HI for both adults and children, with very little additional effect from the  substantial increases in helmet wearing due to the law.

Summary of head injury data

The data in Figs 1-5 show large increases in helmet wearing, but no  major change in %HI, over and above the general trends. These trends  may relate to new diagnostic techniques (e.g. CAT scans), changes in  admission policies (as in SA), and safer roads (leading to lower impact  speeds in collisions, reducing the risk of head injury - Janssen and Wismans, 1985).   Thus it seems impossible to conclude from %HI data  that helmet laws have any large or significant benefit.

Successful road safety measures

Not all road safety measures show almost undetectable  responses.  Road fatalities fell immediately, and remained at  a lower level, when random breath testing was introduced in  NSW (Fig 6).  Some measures – e.g. those encouraging  responsible driving – seem remarkably effective.

In Victoria, campaigns against speeding and drink-driving were  introduced about the same time as the bicycle helmet law.  A  medical journal reported that accident costs were reduced by an  estimated GBP 100M for an outlay of GBP 2.5M (Powles and Gifford, 1993).  Fig 7 shows the fall in  pedestrian fatalities.  Other states also introduced road safety  campaigns about the same time as their helmet laws.  Fig 8 shows  all road casualties in SA in relation to the timing of the helmet law.

Figs 7 and 8 demonstrate why we must take care when claiming  benefits of helmet laws.  Cyclists are likely to benefit just as much as  pedestrians from campaigns to reduce speeding and drink-driving.   Some proponents of helmet laws have shown the equivalent of  Fig 7 and 8 for cyclists, without explaining that similar benefits  were enjoyed by other road users.  The Cochrane Review of  Thompson et al. fails to mention the fall in non-head injuries in  Victoria (Fig 1), and dismisses the much safer road conditions  (Fig 7), leading to the impression that the entire 40% fall in  head cyclists’ head injuries was due to increased helmet  wearing (Thompson, Rivara and Thompson, 2002-9).

Summary of injury data

Despite the lack of obvious change in %HI in response to  increased helmet wearing from legislation (Figs 1-5), proponents of helmet laws have claimed the laws were  effective.  They usually fail to mention important aspects of the data, such as the similar trends in %HI for all  road users (Fig 3), that non-head injuries fell by almost as much as head injuries (Fig 1), or the large  reductions in the amount of cycling (see next section).

Effect on cyclists, health and the environment

Australian and NZ helmet laws are enforced.  In  Victoria, about 20,000 cyclists are fined every year  for not wearing a helmet.  Some cyclists have even  gone to jail for non-payment of helmet-law fines.   Does the threat of a fine encourage cyclists to wear  helmets, or just discourage cyclists who don’t like  helmets from riding?

Table 1 shows the results of a large, comprehensive survey in Melbourne, using the same  observation periods at 64 sites, in similar weather and the same time of year (May).   Before the law, 442 children wore  helmets voluntarily.  A year later, 43  more wore helmets. The big change was  that 649 fewer children were counted.   This strongly suggests that the main  effect of the law, was to discourage  cycling rather than encourage helmet  wearing.  Compared with before the law,  42% few child and 29% fewer adult  cyclists were counted.

Large declines were also noted in a  comprehensive survey of child cyclists at 122 sites covering Sydney, regional and rural areas of NSW. Before  the law, 1910 children were observed wearing helmets. In the first and second years of the law 1019 and 569  more children were observed wearing helmets, but 2215 (36%)  and 2658 (44%) fewer child cyclists were counted (Table 2).

Proponents of helmet laws have argued that the above data  are outdated and distorted by a reduction in the legal driving  age.  This is untrue.  In Victoria, teenagers who pass the driving  test may drive unsupervised from age 18.  This has not  changed, though the minimum age for a learner permit was  lowered from 17 to 16 in Victoria.  Learners must be supervised  at all times by a licensed driver, so it seems unlikely this caused  any significant part of the 42% fall in children’s cycling, or the  29% fall in adult cycling in Melbourne.  Moreover, there was no  change in NSW, yet, by the 2nd year of the helmet law, child  cycle use had fallen by 44%.

There is no evidence that the decline in cycling was transient or that cycling “recovered”.  Fig 9 shows a series of counts over 6  years at 25 sites in Sydney.  Both adult and child cyclists were  counted (Walker, 1996).  There were four surveys in April and two in  October.  More cyclists were observed in April than October,  perhaps because autumn weather may be more conducive to  cycling.  However, by 1996, there were 48% fewer cyclists than  1991.  This is in complete contrast to the situation before the  law, when, in the Sydney metropolitan area "cycling increased  significantly (+250%) in the 1980s" (Webber, 1992).

People often cite helmet laws as a reason for not cycling.  The equivalent of 64% of adult cyclists in Western Australia said they'd ride more except for the helmet law (Heathcote and Maisey, 1994)  In New South Wales, 51% of  schoolchildren owning bikes, who hadn't cycled in the past week, cited helmet restrictions, substantially more  than other reasons, including safety (18%) and parents (20%) (Blacktown).

Long-term trends are also evident in census data on cycling to work.  A comparison of states without  enforced helmet laws in 1991 (Qld,  WA, ACT), and states with  enforced helmet laws, suggests  that the trend of increasing cycling  to work reversed when helmet  laws were enforced. 

Census data for a small area  can be affected by the weather on  census day.  In Melbourne, for  example, the last census before  the helmet law (30 June 2006) had  unusually cold weather (minimum  0.8 C).  Subsequent censuses took  place in August, with census day in  2006 enjoying milder weather  (minimum 5.3 C).  To average out  any local weather effects, the  graph covers the widest possible  area, expressing the total number of people who cycled to work in each group of states as a percentage of  the total number who travelled to work by one mode in those states.

In the mid to late 1980s, cycling was undergoing a surge in popularity. In Western Australia (WA),  numbers cycling more than once a week increased from 300,000 in 1986 to 400,000 (27% of population) in1989. More cycling led to safer cycling – deaths and serious injuries per regular cyclist in WA fell from 5.7 in  1986 to 3.8 in 1989 (Robinson, 1996).  Cycling increased significantly (+250%) in the Sydney metropolitan area (Robinson, 1996).  Long-term  comparisons show substantial declines in cycling compared even to pre-law levels before or during the early  part  of the boom.  Surveys in Western Australia showed a 33% decline in cycling to work from 1986 to 2006  (from about 1.5 to 1.0 trips per weekday per 100 people), consistent with the census data. Shopping trips fell  by 55%, from about 5.2 to 2.3 per weekday per 100 people; trips for education by 79%, from about 8.2 to 1.7  per weekday per 100 people (Ker, 2011).

The Cycling Promotion fund conducted a random survey of 1000 Australians in 2011.  Of 515 respondents  who were not interested in cycling for transport, 81 cited as a reason that they didn’t like wearing helmets.  Only 158 cyclists cycled for transport in the past month - 81 more transport cyclists would represent a 51%  increase, on top of the 16% of current transport cyclists who said that they would cycle more if not required to  wear a helmet (CPF, 2011). A cross-sectional survey in Sydney in 2010 asked is people would cycle more if they did  not have to wear a helmet.  The researchers concluded:

“While a hypothetical situation, if only half of the  22.6% of respondents who said they would cycle more if they did not have to wear a helmet did ride more,  Sydney targets for increasing cycling would be achieved by repealing mandatory bicycle helmet  legislation”. (Rissel and Wen, 2011)

Lack of exercise is a major health problem in most western societies. The New Scientist has reported  that poor diet and physical inactivity may soon overtake tobacco as the leading cause of death.

Other studies show the health benefits of cycling greatly exceed losses from accidents.  UK researcher  Mayer Hillman estimated that 20 life years were gained from the health benefits of cycling (even without a  helmet), for every year lost through accidents (Hillman, 1993). If bicycle helmet laws discourage cycling, as in  Australia, our health will be worse, not better.  If cycling for transport is discouraged, cars may be  used instead, increasing vehicle pollution and greenhouse gas emissions.

Safety in numbers

An interesting paradox is that countries with helmet laws (and therefore high helmet wearing rates) tend to  have higher injury rates per kilometre cycled than countries with low helmet-wearing rates.  This again  suggests that helmet laws are counter-productive.   Helmet law promoters often try to dismiss these facts, but  it is better to examine and understand the evidence, than ignore it.  

A really important paper, published in the prestigious international journal, Injury Prevention, has  a very plausible explanation (Jacobsen, 2003). It reported that the risk per cyclist (and pedestrian) is lower when there are  more cyclists (pedestrians).   For 6 different datasets, a consistent relationship was found, comparing a) data  from different countries, b) different cities and towns in the same country, and c) comparing injury rates and amounts of cycling in the same country over time (Jacobsen, 2003).   This  relationship, called the Growth Rule, shows that when  cycling doubles, injury rates per cyclist falls by an average  of 34%.  Conversely, according to the Growth Rule, if  cycling halves the risk per cyclist will increase by 52% (Jacobsen, 2003).

‘Safety in Numbers’ is also true for Australia.  The last  comprehensive dataset on cycle use was for 1985, but it  shows a very strong relationship between distance cycled  in different Australian states and fatality rates per km  cycled (Fig 10).  States with the most cycling (Australian  Capital Territory, Queensland and Western Australia) had  the lowest fatality rate per km cycled; those with the least  cycling the highest fatality rate per km.(Robinson, 2005b)

 So what happened with the helmet laws?  Did cyclists  have higher injury rates, due to the fall in cycling with the  law?  As noted earlier, we must allow for the fact that helmet laws were introduced at the same time as road  safety campaigns that reduced deaths and serious injuries to other road users (Figs 7 and 8).  Safer roads  should lead to fewer collisions with motor vehicles.  Moreover, when collisions occur, impact speeds should  be lower, reducing the risk of both death and serious head injury.  In experiments designed to mimic  pedestrians and cyclists hit by vehicles, lowering impact speed from 40 to 30 km/hr reduced maximum head  acceleration by 39% (and head injury criterion by 66%), compared to 20-25% reductions in maximum  acceleration of chest, pelvis and feet (Janssen and Wismans, 1985).

We can allow for the safer roads by noting that pedestrian and cyclist injuries follow very similar trends (Robinson, 1996). So, by comparing pedestrian and cyclist safety before and after the law, we can evaluate whether cyclist  head injuries declined (compared to pedestrians) with increased helmet wearing, or whether the fall in cycling  increased injury rates, because of reduced ‘Safety in Numbers’.

Table 3 shows that, in the two years before the law, deaths and serious head injuries (DSHI) represented  26.5% of all serious injuries (ASI) to cyclists in bike/motor vehicle collisions in Victoria.  This fell by 1.7 per- centage points to 24.8% in the 2 years after the law. For pedestrians, the fall over the same period was actu- ally greater – 2.5 percentage points.  Helmets are popularly believed to prevent death and serious head  injury, yet the fall in %DSHI for  pedestrians was actually greater  than that achieved for cyclists  with the law.

Estimated injury rates per  cyclist also suggest that ‘Safety  in Numbers’ operated in reverse.   Pedestrian DSHI fell to 74% of  pre-law numbers (Table 3),  thanks to the road safety campaign already mentioned.  Cyclist DSHI fell to 57% of pre-law numbers, but  there were fewer cyclists – only 69% as many as before the law (Table 1).  DSHI should therefore have fallen  to (69% x 74%) = 52% of pre-law numbers for cyclists to enjoy the same injury reductions as pedestrians.   The actual fall suggests that cyclists did not fare as well with the helmet law as they ought to have done  without it.  Increased injury rates following helmet laws was also noted for child cyclists in NSW (Robinson, 1996). This  strongly suggests that helmet laws are counter-productive and may increase the risk of death and  serious injury.  As well as reduced ‘Safety in Numbers’, another possible factor is risk compensation;  cyclists may take more risks, or motorists take less care when they encounter cyclists, because of  the apparent protection of helmets.

Risk compensation

Although helmets may provide some protection when accidents occur, they may also increase the risk of  accidents because of risk compensation.  Pioneering research in the UK showed that drivers leave less room  when overtaking helmeted cyclists.  The risk of particularly tight passing events doubled when a helmet was  worn. The researcher, Dr Ian Walker, was hit twice - by bus and a truck in the course of the experiment, both  times when wearing a helmet. (Walker, 2007; McLaughlin, 2006)

Other research found that cyclists accustomed to wear helmets ride faster when wearing them, again  suggesting that the act of wearing a helmet makes people feel safer and encourages them to take more risks,  potentially increasing the accident rate. (Phillips, Fyhri and Sagberg, 2011)

Cost-benefit analyses

Any realistic estimate of the cost of helmet laws should include health and environmental costs of reduced  cycling, as well as any increases in injury rates from risk compensation or reduced ‘Safety in Numbers’.  This  has never been done; research into helmet laws has generally been funded by government agencies.  

Nonetheless, the published cost-benefit analyses provide some very interesting information – that the risk  of head injury per cyclist is relatively small.  For this reason alone, helmet laws cost the community money.   For example, in New Zealand, adult cyclists spent NZ$5.9 million on helmets, but the most optimistic estimate  was that the law saved NZD 0.17 million in treating head injuries (See Table 1 of  Taylor and Scuffham, 2002) Even this is debatable.  Head injury data for NZ are shown  in Fig 2.  Helmet wearing of adults rose dramatically with the law, but there was little change for primary schoolchildren.  Yet percent  head injury (%HI) of both groups follow similar trends.  In fact, %HI for primary schoolchildren actually declined slightly more than adults.  Thus, despite its $5.9 million price tag, the helmet law for adults may not  have prevented any head injuries whatsoever. The most optimistic estimate of the savings in head injury treatment costs (Table 4) is NZD 0.13 per helmet per year.

Table 4. NZ helmet law - most optimistic (MO) estimates of the savings in healthcare costs (from Taylor and Scuffham, 2002)

Age group 5-12 13-18 Adult All
Number of helmets required 10,195 84,999 328,162 423,356
Helmet cost (NZD million, per 5 years) 0.47 1.28 4.29 6.04
MO estimates - head injury treatment saving (NZD million, per 5 years) 0.03 0.08 0.17 0.28
MO estimates - head injury treatment saving per helmet per year (NZD) 0.56 0.18 0.11 0.13

Another cost-benefit analysis was carried out for Western Australia (WA) by Delia Hendrie, University of  WA.  She estimated that the WA helmet law cost $21.6 million, including $20.2 million to purchase helmets (Hendrie, Legge, Rosman and Kirov, 1999).As discussed earlier, %HI of cyclists followed similar trends to that of other road users (Fig 3).  However, the  graph has a very curious feature – a large increase in %HI of pedestrians (but not other road users) in year  21, a year before the helmet law.  This seems to have been transient.  By year 25, pedestrian %HI had fallen  below that of cyclists.  However, if, without the law, cyclists %HI were presumed to follow exactly the same  trend as pedestrians (a highly contentious assumption!) the average difference over this period, about 20 to  44 head injuries per year, mostly of moderate severity, might (albeit unrealistically) be deemed an effect of  the law.  Depending on the model fitted and how costs of head injuries are calculated, net benefits, estimated  in this unrealistic way, range from AUD -15.1 million (i.e. the law still cost AUD 15.1 million more than an  unrealistically optimistic estimate of what it saved) to AUD 2 million.

Recent attention has focused on the health benefits of cycling.  A Danish study reported that, after  adjustment for age, sex, and educational level, people who did not cycle to work had about 40% higher  mortality than those who did. Adjusting for leisure time physical activity, body mass index, blood lipid levels,  smoking, and blood pressure made very little difference – not cycling to work increased mortality rates by  39%. (Andersen, Schnohr, Schroll and Hein, 2000)

An economic analysis by Price Waterhouse Coopers for the NSW RTA in 2009 concluded: “Using the  most conservative assumptions in all cases, the estimated net benefits of cycling were found to be 48.22  cents per bicycle kilometer” (PWC, 2009). This again suggests that the benefits of cycling are much greater than the  estimated maximum reduction in hospital costs (NZ data, 0 to 13 cents per year, Table 4 above.) 

A Canadian researcher also explored the costs and benefits of bicycle helmet laws, concluding that if the  helmet law reduces cycling by more than a quarter of one per cent, there will be a net loss, rather than a  benefit.  It was estimated that British Columbia’s helmet law has cost 500 lives since its introduction in 1996,  compared to an estimated saving of 7 lives. (Johns, 2011)

Thus if estimates of lost benefits from reduced cycling are included in the costs, helmet laws cost  far more than any realistic estimate of their benefits.

What causes brain injury?

Experimental evidence shows that brain damage is caused mainly by rotation.  Rotational forces can  shear the brain’s neuronal connections, a condition known as diffuse axonal injury (DAI).  In one experiment  12 squirrel monkeys were subjected to linear accelerations with peak levels 665-1230 g, and 13 primarily to  rotational accelerations in the range of 348 to 1025 g (Gennarelli, Thibault and Ommaya, 1972). Contact phenomena were minimised by the design  of the apparatus.  None of the monkeys receiving linear acceleration was concussed, but all 13 receiving  rotational acceleration suffered concussion, and the group had a high incidence of brain injuries such as  subdural haematoma, subarachnoid haemorrhage and intracerebral petechial haemorrhage.

There is little reliable evidence whether bicycle helmets reduce, or increase, the risk of rotational injury.   No-shell helmets may stick on pavement and increase the risk of rotating the head.  The Australian NHMRC  (National Health and Medical Research Council) discussed football helmets (which may have some similarity  with bike helmets), stating: 

"The use of helmets increases the size and mass of the head. This may result in an  increase in brain injury by a number of mechanisms.  Blows that would have been glancing become more solid and thus  transmit increased rotational force to the brain. These forces result in shearing stresses on neurones which may result in  concussion and other forms of brain injury." (NHMRC, 1994)

Does this happen in real life?  In a paper, "Cycling: your health, the public’s health and the planet’s  health", Public Health Physician, Dr Ashley Bloomfield wrote:

"The earliest murmurings that I heard against  helmets were from a neurosurgeon who I worked for in 1994. He claimed that cycle helmets were turning what would  have been focal head injuries, perhaps with an associated skull fracture, into much more debilitating global head  injuries. We had a couple of examples on the ward at the time, and it was a bit worrying. However, I wasn’t too  convinced as I figured that the injuries that would previously have been focal head injuries may well have been resulting  in death, so the neurosurgeon was never actually seeing them. Instead, they were making their way straight to the  pathologist." (Bloomfield, 2000)

However, the comparison of pre- and post-helmet law statistics for pedestrians and cyclists in Victoria  (Table 3), suggests that helmets prevent few, if any deaths or serious head injuries (DSHI).  In fact, DSHI to  cyclists, as a proportion of all serious or fatal injuries to cyclists (%DSHI), fell by (marginally) less than the  same statistic for pedestrians.  More importantly, the risk of death or serious head injury per cyclist actually  increased compared to what would have been predicted without the law.  

Thus, there is some worrying evidence that helmets might increase the risk of rotational injuries.  This may  explain the increase in DSHI, compared to what would have been expected without the law. Other factors  such as risk compensation and reduced ‘Safety in Numbers’ may also be important.  Available data are  insufficient to draw any firm conclusions.  One problem is that diffuse axonal injury (DAI) is very hard to see  on CAT scans, so many cases may be missed.

In motorcyclists, the serious brain damage from rotational injuries (despite, presumably, wearing traditional  helmets) was noted by physician Dr Ken Phillips.  He was so concerned he decided to try and improve  matters.  Dr Phillips observed that the scalp provides the brain with protection against rotational forces  because it is elastic, compressible and moves around the skull without friction.  To mimic this process, the  'Phillips helmet' has an outer shell of polyethylene that moves independently of the inner cushion (Phillips helmets).  Unfortunately, no helmet manufacturer was interested,  so Dr Phillips started his own company to make and market them.

The final paradox

The evidence reported here indicates that helmet laws are not effective and may even increase the risk of  injury per cyclist.  The many reasons for this include: risk compensation, reduced safety in numbers, failure to wear helmets correctly and the fact that helmets probably offer little protection against the most debilitating  type of brain injury – rotational injuries.

So why is so much effort expended promoting helmet laws?  One source of confusion is that case-control  studies comparing cyclists choosing to wear helmets with helmetless cyclists usually show that helmet  wearers have a lower %HI.  But why?  Is this because helmets offer some benefit, or that cyclists choosing to  wear helmets are more cautious, have different riding styles and get into less serious accidents?  Studies  show that cyclists who chose to wear helmets are more likely to obey traffic signs, wear high visibility  clothing, have higher socioeconomic status, use lights at night, ride in parks, playgrounds, or on bicycle  paths, rather than city streets, (in the US) be white rather than other races, and (for children) tend to ride with  other children or adults, rather than alone (Robinson, 1996). One US study also found that helmet wearers had much less  serious non-head injuries, as well as head injuries. 

Although case-control studies try to control for these differences (known as confounding factors), it is  virtually impossible to record and control for all differences between wearers and non-wearers.  This problem  was discussed in a series of papers in the International Journal of Epidemiology (IJE) in June 2004, after a  similar paradox was noted for hormone replacement therapy (HRT). (Petitti, 2004)

In 1991, a review of the best quality observational studies concluded that HRT reduced the risk of coro- nary heart disease by 50%, and that ‘overall, the bulk of the evidence strongly supports a protective effect of  estrogens that is unlikely to be explained by confounding factors’ (Lawlor, Smith and Ebrahim, 2004). Yet when randomized control trials were  carried out, the exact opposite was found – that HRT increased the risk of coronary heart disease by 29%.   Just as cyclists who choose to wear helmets tend to be more cautious and have higher socioeconomic  status, women choosing HRT also tend to have higher socioeconomic status.  Results from the observational  studies had, in fact, suggested the possibility of bias from confounding; HRT was apparently equally  protective against accidental and violent deaths as it was against cardiovascular disease deaths in one  observational study. (Lawlor, Smith and Ebrahim, 2004)

Can similar checks for bias be applied to case-control studies of helmet efficacy?  In the most widely citied  study, most (86%) of the community controls were children under 15 (Thompson, Rivara and Thompson, 1989). 21.1% wore helmets when they fell  of their bikes, compared to 5.9% of 202 children given emergency room treatment in Seattle for non-head  injuries.  The odds ratio for helmets preventing head injury (HI) was calculated by comparing the 143 HI  children (‘cases’) with the community controls (Thompson, Rivara and Thompson, 1989). In exactly the same way, children treated for injuries to  other parts of the body can be considered ‘cases’.  This produces an odds ratio of 0.23, indicating that  helmets prevent 77% of injuries to other parts of the body!  This suggests, as was found for the HRT  observational studies, that much of the estimated benefits could be artifacts of confounding.  

Another indicator of potential confounding was that surveys at the same time as the widely-cited Seattle  study found that 3.2% of 4501 child cyclists riding round Seattle wore helmets (DiGuiseppi, Rivara, Koepsell and Polissar, 1989). This is not significantly  different from the 2.1% and 5.9% helmet wearing (%HW) of children with head and non-head injuries.   However, all three are completely different from the 21.1% HW of the community controls (p < 0.001).  We  might therefore conclude that helmet wearers are nearly 7 times more likely to fall off their bikes than non- wearers, suggesting that risk compensation is far more important than any possible benefits of helmets.   However, this may also be an artifact of confounding.  Community controls were members of a Group  Healthcare Cooperative that may have promoted helmet wearing to members.  Children might also be  persuaded to wear helmets when they are more likely to fall off their bikes, e.g. when learning to ride.  Thus we may conclude it is very difficult to fully adjust for confounders in observational studies of helmet efficacy.  

The best estimates of the benefits of helmets laws are those based on what actually happens when  such laws are introduced.  As shown in Figs 2-5, even though other road safety measures achieved  large reductions in injury rates (Figs 6-8), there were no major changes in cyclists %HI when helmet  laws were introduced.  In contrast, the large fall in non-head injuries (Fig 1) suggests the laws  discouraged cycling.


Comparisons of pre- and post-law injury data (Figs 1-5) show that there is little benefit to either cyclists or  the community from passing laws forcing cyclists to wear helmets.  Rather than encouraging cyclists to wear  helmets, the laws appear to have discouraged cycling, resulting in reduced health and fitness, but very little  change in %HI. If the money spent on helmets had been used for other measures e.g. improving accident  blackspots for cyclists, the benefits would have much been greater.

More importantly, risks per cyclist seem to have increased, compared to what would have been expected  without the law, implying that helmet laws are counter-productive.  Possible explanations include risk  compensation, reduced ‘Safety in Numbers’ and that brain damage is predominantly due to rotational injury.  

Helmets undoubtedly help prevent minor wounds to the head, but are not designed to cope with the forces  that may occur in bike/motor vehicle collisions.  There is little reliable experimental evidence whether bicycle helmets reduce, or increase, the risk of rotational brain injury.  However, as a precaution, cyclists choosing to  wear helmets may wish to consider new designs such as the ‘Phillips’ helmet.

In contrast to the little or no obvious change in %HI with bicycle helmet laws, injury statistics following  measures to reduce speeding and drink driving (Figs 6-8) show considerable benefit.  The vast majority of  fatal and very serious head injuries to cyclists result from bike/motor vehicle collisions. The most effective  way to reduce injuries to cyclists and all other road users is therefore to reduce the risk of bike/motor vehicle  collisions.  

As well as enforcing appropriate speed limits, controlling drink-driving and encouraging cyclists to use  lights at night and ride on the correct side of the road, cycling becomes safer when more people cycle.  The  best option to improve overall safety, improve our health and fitness and benefit the environment is therefore  a package of measures to encourage cycling and make the roads safer, while allowing cyclists to chose  whether of not they wish to wear helmets.

Sources of data

The data shown have been compiled over many years from many sources.  Data on injuries to cyclists and  pedestrians in Victoria (Table 3) were kindly provided by the Victorian Transport Accident Commission. The appendix  provides more details on how head injuries are classified.  Fatality data (Figs 6 and 7) were obtained from the ATSB  website ( Other data were obtained from the referenced papers and reports.

Appendix: Other published analyses and description of head injuries

Other published analyses of head injury data

The data in Figs 1-5 show that, in contrast to the large benefits from campaigns against speeding and drink-driving  (Fig 6-8), there was no obvious change in %HI from bicycle helmet laws.  Indeed, the large reduction in non-head  injuries (Fig 1) suggests the main effect of the laws was to discourage cycling, rather than protect against head injury.   Despite this, proponents of helmet laws claim they are effective; they advise cyclists that wearing a helmet may save  their lives.  This divergence of opinion is discussed by reviewing research frequently cited by advocates of such laws.

One widely-cited report is: Carr et al. (Monash Univ Accident Research Centre (MUARC), 1995) (Carr, Dyte and Cameron, 1995).  It notes (as is  obvious in Fig 1) that numbers of head injuries fell by 40% after the helmet law in Victoria.  The authors explained they  could not tell from their analysis whether this was because fewer people cycled after the law, or because helmet wearing  increased.  So why analyze numbers of head injuries?  MUARC’s analysis of the first 3 years of helmet law data found  that %HI was no different to that predicted from pre-law trends (Cameron, Newstead, Vulcan and Finch, 1994).   Might we speculate the main reason for focussing on  numbers (rather than percentages) of head injuries was a desire to justify the helmet law?

Another widely-cited publication was MUARC’s analysis of the first 2 years of helmet-law data:

"(TAC) insurance  claims from bicyclists killed or admitted to hospital after sustaining a head injury decreased by 48% and 70% in the first  and second years after the law, respectively. Analysis of the injury data also showed a 23% and 28% reduction in the  number of bicyclists killed or admitted to hospital who did not sustain head injuries in the first and second post-law  years, respectively." (Cameron, Vulcan, Finch and Newstead, 1994)

Fig 1 shows all cyclist hospital admissions, irrespective of whether the cyclist fell off the bike, or was hit by a fast  moving motor vehicle.  Although risk compensation might also be important, we may hypothesize that the dominant  influence on numbers of non-motor-vehicle injuries is the amount of cycling.  The observational surveys (Table 1), and  the fall in numbers of non-head injuries (Fig 1), strongly suggest the main effect of the law was to discourage cycling.   Thus the most likely cause of the 40% reduction in head injuries was that the law discouraged cycling.

In contrast, insurance claims for injuries in motor vehicle accidents also depend on other factors, including driver  behaviour and vehicle travelling speeds.  This was demonstrated by the 42% in pedestrian deaths from1989 to 1990  (Fig 7).  Table 3 compares deaths and serious head injuries (as a percentage of all serious injuries, %DSHI) of  pedestrians and cyclists in motor-vehicle accidents.  The fall %DSHI of pedestrians was actually greater than for  cyclists.  Indeed, the risk of DSHI per cyclist appears to have increased relative to that for pedestrians.  

MUARC didn’t use pedestrians as a control and didn’t explain that numbers of pedestrians with concussion fell by  29% and 75% in the first and second years of the helmet law.  This reduction for pedestrians is almost as impressive as  the 48% and 70% falls in head injuries of cyclists over the same period, compared to 23% and 28% falls in other injuries.   It implies that helmets were not the main cause.  One refereed paper set the record straight by comparing all head  injuries, including concussion, of pedestrians and cyclists, showing both followed almost identical trends. (Robinson, 1996)

Section 3 uses the alternative approach of considering only deaths and injuries serious enough to warrant admission  to hospital.  Table 3 was based on TAC’s classification of serious head injury, i.e. skull fracture (ICD-9 codes 800, 801,  803, 804) and brain injuries (codes 851-854), but not or wounds to the head or concussion with no other sign of brain  injury.  This contrasts with MUARC’s definition of "bicyclists killed or admitted to hospital after sustaining a head injury",  which included concussion and wounds to the head or ears, even if the main reason for admission was a serious injury  to another part of the body. (Cameron, Vulcan, Finch and Newstead, 1994)

Table 3 shows the risk of death and serious head injury (as defined by TAC) increased relative to that for  pedestrians.  We may therefore conclude that the widely-quoted claims of 48% and 70% reduction in cyclist head  injuries were mainly due either to changes in admission procedures and improved road safety conditions (resulting in  29% and 75% reductions in numbers of concussions to pedestrians).  Cyclists admitted to hospital for treatment of  serious injuries to other parts of the body may also have had fewer head wounds.  Most people wear helmets because  they hope to be protected against death and serious head injury.  Despite the claims of helmet law proponents, TAC  data suggest that the helmet laws in Victoria actually increased the risk of DSHI relative to the amount of cycling.

Other published analyses of the amount of cycling

The abstract of MUARC publication claiming 48% and 70% reductions in head injuries also stated:

“Surveys in  Melbourne also indicated a 36% reduction in bicycle use by children during the first year of the law and an estimated  increase in adult use of 44%”

Yet the survey data (Table 1) show 29% fewer adults were counted in May 1991 than May 1990.  So why did  MUARC claim that adult cycling increased by an estimated 44%?  MUARC estimated cycle use from the time cyclists  took to ride through marked areas.  However, in 1990, adult cyclists were counted, but not timed.  This should not  preclude adult cycle use from being estimated, because numbers counted and estimates of cycle use are strongly correlated.  For example, the first post-law survey found a decline in cycle use by teenagers of 44%, little different from  the 48% drop in numbers counted.  This implies that the 29% decline in number of adults counted is a valid and  reasonable estimate of the change in adult cycle use.  

But instead of reporting this direct estimate, MUARC ignored the number of adults counted in 1990, and estimated  the effect of the law by comparing adult cycle use in 1991 with a much earlier survey (1987/88) at a different time of  year.  This is a totally invalid because (as can be seen in Figure 1) cycle use has a marked seasonal variation.  If the  same “trick” of ignoring data from1990 had been carried out for teenagers, MUARC could have claimed the law reduced  teenage cycle use by a mere 8%, instead of the 44% actually observed!

The decisions taken by MUARC to base their final analysis on numbers of head injuries, rather than percentages,  and ignore numbers of adult cyclists counted in 1990, created considerable confusion about the effects of helmet laws.   Governments may have wished the laws to be considered successful.  However, readers of the original reports would  understand the data were open to other interpretations.  Unfortunately, summary reports by proponents of helmet laws  in other countries often omitted all-important details, leading to the false impression that the laws were remarkably  successful.  Figs 1-5, showing the original data, demonstrate that this was not the case.  Readers should inspect the  graphs, in conjunction with details about how the data were compiled (below), and make up their own minds. 

Mistakes in relation to HRT were discussed in the International Journal of Epidemiology.  Four important lessons had  to be learned: 1) do not turn a blind eye to contradiction, 2) do not be seduced by mechanism, 3) suspend belief and 4)  maintain skepticism (Petitti, 2004). Bicycle helmet researchers, who use same methods, need to learn the same lessons.  Helmet  law promoters believe helmets work; the mechanism (they absorb energy) is seductive.  Cyclists who doubt the benefits  may be less likely to wear helmets.  So promoters ignore contradictory evidence about risk compensation, causes of  brain injury and that increased wearing from 30-40% to 75-90% following helmet laws produced no large or obvious  response in %HI.   

Helmet laws promoters also cite data showing generally declining trends in %HI coinciding with increasing trends in  helmet wearing.  But Fig 3 shows declining trends commonly affect all road users.  It is therefore impossible to separate  effects of long-term trends from gradual changes in helmet wearing as a result of education, or non-enforced laws.  

The Australian and New Zealand data are unique in that they produced very large, rapid, increases in helmet wearing  – 40 or 50 percentage points.  Yet there was no obvious response in %HI, but large reductions in the amount of cycling,  (Figs 1-5, Tables 1-3).  This provides strong, compelling evidence that helmet laws do not work, refuting the claims of  those who wish to compel all cyclists to wear helmets.

Classification of head injury data (Figs 1 - 5, Table 3)

As shown in the acknowledgments, Figs 1-5 were complied from other publications, using other people’s  classification of head injury.  The example for Victoria (Appendix A1) showed that different classifications may reveal  different effects. No effect of the helmet law was detectable in comparison with pedestrians, but larger trends in %HI  were evident when concussion (with no other sign of brain injury) was included.  The definitions of head injury for Figs 1- 5 and Table 3 are therefore listed below to enhance understanding of the data.

Most classifications included all head injuries: skull and facial fractures, intracranial injuries, concussion and wounds  to the head.  As discussed in Section 5, the most debilitating brain injuries may be caused by rotations, for which  helmets might not offer any protection.  However, helmets undoubtedly prevent wounds to the head, so some effect of  increased helmet wearing could be expected when head wounds are included – Figs 1 (Vic), 3 (WA) and presumably 5  (NSW), but not Fig 2 (NZ).  Fig 4 (SA) used principle diagnosis, so a decrease in minor wounds to the head of cyclists  admitted to hospital for other reasons may not be apparent. 

For WA, Fig 3 shows data from 1971-95, classified by Hendrie, Legge, Rosman and Kirov, 1999  for cyclists, pedestrians and motor vehicle  occupants.  An alternative classification, only for cyclists, from 1981-95, based on principle diagnosis, was also  published – see   Data for skull fractures and intracranial  injuries were very highly correlated (r=0.98) with the cyclist data in Fig 3.  A much better insight into the full picture is  provided by Fig 3, because it also shows the same statistics for other road users.


Andersen, Schnohr, Schroll and Hein, 2000

Andersen LB, Schnohr P, Schroll M, Hein HO, 2000. All-cause mortality associated with physical activity during leisure time, work, sports, and cycling to work. Arch Intern Med 2000 Jun 12;160(11):1621-8.


Blacktown Bikeplan Study, Final Report. Blacktown City Council, Sydney.

Bloomfield, 2000

Bloomfield A, 2000. Cycling: your health, the public's health and the planet's health. Making Cycling Viable symposium .

Cameron, Newstead, Vulcan and Finch, 1994

Cameron M, Newstead S, Vulcan P, Finch C, 1994. Effects of the compulsory bicycle helmet wearing law in Victoria during its first three years. Monash University Accident Research Centre .

Cameron, Vulcan, Finch and Newstead, 1994

Cameron MH, Vulcan AP, Finch CF, Newstead SV, 1994. Mandatory bicycle helmet use following a decade of helmet promotion in Victoria, Australia - an evaluation. Accident Analysis & Prevention 1994;26(3):325-337.

Carr, Dyte and Cameron, 1995

Carr D, Dyte D, Cameron MH, 1995. Evaluation of the bicycle helmet wearing law in Victoria during its first four years. Monash University Accident Research Centre Report 76.

CPF, 2011

Riding a bike for transport - 2011 survey findings. Canberra: Cycling Promotion Fund. 2011. .

DiGuiseppi, Rivara, Koepsell and Polissar, 1989

DiGuiseppi CG, Rivara FP, Koepsell TD, Polissar L, 1989. Bicycle helmet use by children. Evaluation of a community-wide helmet campaign. J Amer Med Assoc 262:2256-61.

Gennarelli, Thibault and Ommaya, 1972

Gennarelli TA, Thibault LE, Ommaya AK, 1972. Pathophysiological responses to rotational and translational accelerations of the head. Society of Automotive Engineers Paper 720970.

Heathcote and Maisey, 1994

Heathcote B, Maisey G, 1994. Bicycle use and attitudes to the helmet wearing law. Traffic Board of Western Australia May 1994.

Hendrie, Legge, Rosman and Kirov, 1999

Hendrie D, Legge M, Rosman D, Kirov C, 1999. An Economic Evaluation of the Mandatory Bicycle Helmet Legislation in Western Australia. Road Accident Prevention Research Unit .

Hillman, 1993

Hillman M, 1993. Cycle helmets: the case for and against. Policy Studies Institute ISBN 0-85374-602-8.

Jacobsen, 2003

Jacobsen PL, 2003. Safety in numbers: more walkers and bicyclists, safer walking and bicycling. Injury Prevention 2003;9:205-209.

Janssen and Wismans, 1985

Janssen EG, Wismans JSHM, 1985. Experimental and mathematical simulation of pedestrian-vehicle and cyclist-vehicle accidents. 10th International Tech Conference on Experimental Safety Vehicles Oxford July 1985.

Johns, 2011

Johns R, . Cost Benefit Analysis. How it works, and when it doesn’t work. . University of British Columbia, 2011. Includes consideration of helmet benefits..

Ker, 2011

Ker I, . Empty Cells, Damned Half-Truths and Pseudo-Statistics:The Lot(tery) of the Bicycle Planner. Paper presented at the the PATREC Planning and Transport Research Centre Forum, 13 September 2011.

Lawlor, Smith and Ebrahim, 2004

Lawlor DA, Smith GD, Ebrahim S, 2004. The hormone replacement - coronary heart disease conundrum: is this the death of observational epidemiology?. Int Journal of epidemiology 2004;33:464-467.

Marshall and White, 1994

Marshall J, White M, 1994. Evaluation of the compulsory helmet wearing legislation for bicyclists in South Australia. South Australia Dept of Transport Report 8/94.

McLaughlin, 2006

Wearing a helmet puts cyclists at risk, suggests research. 2006.

NHMRC, 1994

Football injuries of the head and neck. National Health & Medical Research Council, AGPS, Canberra .1994.

Petitti, 2004

Petitti D, 2004. Hormone replacement therapy and coronary heart disease: four lessons. Int Journal of Epidemiology 2004;33:461-463.

Phillips helmets

Phillips helmets. .

Phillips, Fyhri and Sagberg, 2011

Phillips RO, Fyhri A, Sagberg F, 2011. Risk compensation and bicycle helmets. Risk Analysis 2011; Mar 18.

Powles and Gifford, 1993

Powles JW, Gifford S, 1993. Health of nations: lessons from Victoria, Australia. BMJ 1993 Jan 9;306(6870):125-7.

PWC, 2009

Evaluation of the costs and benefits to the community of financial investment in cycling programs and projects in New South Wales.. Price Waterhouse Coopers, 2009.

Rissel and Wen, 2011

Rissel C, Wen LM, 2011. The possible effect on frequency of cycling if mandatory bicycle helmet legislation was repealed in Sydney, Australia: a cross sectional survey. Health Promotion Journal of Australia 2011; 22: 178-83.

Robinson, 1996

Robinson DL, 1996. Head injuries and bicycle helmet laws. Accident Analysis & Prevention 1996 Jul;28(4):463-75.

Robinson, 2001

Robinson DL, 2001. Changes in head injury with the New Zealand bicycle helmet law. Accident Analysis & Prevention 2001 Sep;33(5):687-91.

Robinson, 2005b

Robinson DL, 2005. Safety in numbers in Australia: more walkers and bicyclists, safer walking and bicycling. Health Promotion Journal of Australia 2005;16:47-51.

Taylor and Scuffham, 2002

Taylor M, Scuffham P, 2002. New Zealand bicycle helmet law - do the costs outweigh the benefits?. Injury Prevention 2002;8:317-320.

Thompson, Rivara and Thompson, 1989

Thompson RS, Rivara FP, Thompson DC, 1989. A case control study of the effectiveness of bicycle safety helmets. New England Journal of Medicine 1989 v320 n21 p1361-7.

Thompson, Rivara and Thompson, 2002-9

Thompson DC, Rivara FP, Thompson RS., 2002. Helmets for preventing head and facial injuries in bicyclists (Cochrane Review). Cochrane Database Syst Rev issue 4, 2002.

Walker, 1996

Walker M, 1996. Bicycling in Sydney: law compliance and attitudes to road safety. Velo Australis .

Walker, 2007

Walker I, 2007. Drivers overtaking bicyclists: Objective data on the effects of riding position, helmet use, vehicle type and apparent gender. Accident Analysis & Prevention 2007 Mar;39(2):417-25.

Webber, 1992

Webber R, 1992. Cycling in Europe. Ausbike 1992 proceedings, Bicycle Fed of Australia .