It has been suggested that seat-belt legislation may result in drivers' feeling overprotected and their consequent riskier driving may lead to more injuries to other road users (the risk homeostasis or compensation theory). We have examined data in Britain before and after legislation made seat-belt use compulsory for front-seat occupants of specified vehicles. The relative numbers of injuries to cyclists and pedestrians in collisions with vehicles requiring and not requiring seat-belt use were compared, using a log-linear analysis taking into account possible confounding variables. It was not possible directly to incorporate vehicle distance travelled for the two categories of vehicle. However, a simple adjustment for differential vehicle use accounted for most of an apparent effect. The revised odds ratio (for mandated seat-belt use vehicles versus nonmandated, post- versus prelegislation) was 1.04 for cyclists and 1.06 for pedestrians, showing minimal, if any, evidence for the theory.