Bicycle helmet efficacy was quantified using a formal meta-analytic approach based on peer-reviewed studies. Only those studies with individual injury and helmet use data were included. Based on studies from several countries published in the period 1987-1998, the summary odds ratio estimate for efficacy is 0.40 (95% confidence interval 0.29, 0.55) for head injury, 0.42 (0.26, 0.67) for brain injury, 0.53 (0.39, 0.73) for facial injury and 0.27 (0.10, 0.71) for fatal injury. This indicates a statistically significant protective effect of helmets. Three studies provided neck injury results that were unfavourable to helmets with a summary estimate of 1.36 (1.00, 1.86), but this result may not be applicable to the lighter helmets currently in use. In conclusion, the evidence is clear that bicycle helmets prevent serious injury and even death. Despite this, the use of helmets is sub-optimal. Helmet use for all riders should be further encouraged to the extent that it is uniformly accepted and analogous to the use of seat belts by motor vehicle occupants.