As yet there has not been reported an adequate field demonstration of the protective effects of bicycle safety helmets. Evaluation of helmet efficacy has rested almost entirely on laboratory tests of questionable relevance to real world accidents. In the present study 894 South Australian cycling enthusiasts were contacted by mail with regard to their most recent accident and their helmet use at the time. Overall, 197 bicyclists were identified who had experienced an accident within the past 5 years, and had struck their head or helmet in the accident. Helmet status ranged from no helmet used (n = 75), to hairnetstyle helmet (n = 69), to unlined solid helmet (n = 37), to lined solid helmet (n = 16). Analysis of the crude, unadjusted data showed a statistically significant association between helmet use and reduced severity of head injury. The association persisted after adjustment for age, sex and severity of crash forces. It is estimated that the risk of death from head injury is 3-10 times greater for unhelmeted relative to helmeted bicyclists, depending on helmet type.