A systematic review of the literature was undertaken to evaluate the criteria used for the diagnostic classification of traumatic dental injuries from an epidemiological standpoint. The methodology used was that suggested by the Cochrane Collaboration and the National Health Service. A total of 12 electronic bibliographical databases (BBO, BioMed Central, Blackwell Synergy, Cochrane, DARE, EMBASE, HighWire, LILACS, MEDLINE, PubMed Central, SciELO, SciSearch) and the World Wide Web were searched. There was no attempt to specify the strategy in relation to date, study design, or language. The last search was performed in May 2003. Two reviewers screened each record independently for eligibility by examining titles, abstracts, keywords and using a standardized reference form. Disagreements were resolved through consensus. The final study collection consisted of 164 articles, from 1936 to 2003, and the population sample ranged from 38 to 210,500 patients. 54 distinct classification systems were identified. According to the literature, the most frequently used classification system was that of Andreasen (32%); as regards the type of injury, the uncomplicated crown fracture was the most mentioned lesion (88.5%). Evidence supports the fact that there is no suitable system for establishing the diagnosis of the studied injuries that could be applied to epidemiological surveys.