We conducted a systematic review to assess the evidence for the effectiveness of stretching as a tool to prevent injuries in sports and to make recommendations for research and prevention. Without language limitations, we searched electronic data bases, including MEDLINE (1966-2002), Current Contents (1997-2002), Biomedical Collection (1993-1999), the Cochrane Library, and SPORTDiscus, and then identified citations from papers retrieved and contacted experts in the field. Meta-analysis was limited to randomized trials or cohort studies for interventions that included stretching. Studies were excluded that lacked controls, in which stretching could not be assessed independently, or where studies did not include subjects in sporting or fitness activities. All articles were screened initially by one author. Six of 361 identified articles compared stretching with other methods to prevent injury. Data were abstracted by one author and then reviewed independently by three others. Data quality was assessed independently by three authors using a previously standardized instrument, and reviewers met to reconcile substantive differences in interpretation. We calculated weighted pooled odds ratios based on an intention-to-treat analysis as well as subgroup analyses by quality score and study design. Stretching was not significantly associated with a reduction in total injuries (OR = 0.93, CI 0.78-1.11) and similar findings were seen in the subgroup analyses. There is not sufficient evidence to endorse or discontinue routine stretching before or after exercise to prevent injury among competitive or recreational athletes. Further research, especially well-conducted randomized controlled trials, is urgently needed to determine the proper role of stretching in sports.