Accident related health problems have been suggested to cluster within persons. This phenomenon became known as accident proneness and has been a subject of many discussions. This study provides an overview of accident proneness. Therefore, 79 articles with empirical data on accident rates were identified from databases Embase, Medline, and Psychinfo. First, definitions of accidents varied highly, but most studies focused on accidents resulting in injuries requiring medical attention. Second, operationalisations of accident proneness varied highly. Studies categorised individuals into groups with ascending accident rates or made non-accident, accident, and repetitive accident groups. Third, studies examined accidents in specific contexts (traffic, work, and sports) or populations (children, students, and patients). Therefore, we concluded that no overall prevalence rate of accident proneness could be given due to the large variety in operationalisations. However, a meta-analysis of the distribution of accidents in the general population showed that the observed number of individuals with repeated accidents was higher than the number expected by chance. In conclusion, accident proneness exists, but its study is severely hampered by the variation in operationalisations of the concept. In an effort to reach professional consensus on the concept, we end this paper with recommendations for further research. (C) 2006 Elsevier Ltd. All rights reserved.