The paper describes an evaluation of impact performance requirements for pedal cycle helmets. The paper examines the results of two related studies, evaluates other helmet test results and proposes performance criteria more effective for the amelioration of head injury. The two main studies are of pedal cycle helmet performance in real accidents (McIntosh and Dowdell IRCOBI 1992) and head impact tests conducted under conditions relevant to those occurring during pedal cycle accidents (McIntosh et al Stapp 1993). The results of other helmet evaluations are drawn upon. The paper examines a number of areas of helmet performance and focuses on head coverage and impact test criteria.The results of the studies demonstrate that pedal cycle helmets are failing to provide adequate coverage in the temporal region, and that standards tests are not sensitive to this problem. The Test Line could be measured directly from the Frankfort Plane delineating a region below which head impacts are unlikely or do not pose a serious risk of brain injury. The helmet should cover the head almost to the Frankfort Plane laterally and posteriorly.Although drop heights of 1 500 mm result in a suitable impact test velocity, the permissible acceleration levels are too high. Statistical analysis of head impact data showed that maximum acceleration was a suitable test criterion for helmet performance. but that pass/fail criteria should be reduced, i.e. maximum centre of gravity head acceleration < 200 g. Levels of acceleration of 300 to 400 g, currently allowed by helmet standards are too high.