This paper aims to review and collate the epidemiological data of injuries in competition taekwondo as reported in the literature, make recommendations, and suggest further research. The electronic databases AMED, AusportMed, CINAHL, MEDLINE, PubMed, and SPORTDiscus were searched from inception to March 2008. Fourteen prospective cohort studies reporting the incidence of injuries in taekwondo were included. Two reviewers independently extracted data and assessed trial quality using the STROBE statement. Homogenous studies were combined in a pooled analysis using a Poisson random effects regression model. Poisson regression showed an overall mean injury rate of 79.3 per 1000 athlete-exposures (95% confidence interval 22.8, 275.4). Neither age, gender nor level of play were significant in the analysis. The most common injury location and type were found to be the lower limb and contusion, respectively, and were invariably associated with contact. Although taekwondo players are exposed to a substantial risk of sustaining injuries, the majority of injuries appeared to be of minimal severity. Modifications to the competition rules and protective equipment may be warranted. Future studies should adhere to recommended operational definitions, utilise a standardised injury classification system, and report injury rates using multiple denominators in order to facilitate inter- and cross-sport comparisons.