PURPOSE: Information on the potential relation between marijuana use and the incidence of hospitalized injury is extremely limited. The purpose of this effort was to investigate the potential for this association. METHODS: A retrospective study was conducted in a large prepaid Northern California health care program cohort (n = 64,657) that completed baseline questionnaires about health behaviors, including marijuana use, and health status between 1979 and 1985. All injury hospitalizations through December 31, 1991, (n = 965) were identified and validated. RESULTS: Using Poisson regression modeling, increased rate-ratios and 95% confidence intervals were identified for all-cause injury hospitalizations for both men and women among current users (1.28; 1.01 to 1.61 and 1.37; 1.04 to 1.79, respectively) relative to nonusers, adjusted for age, cigarette and alcohol use, and other potential confounders. Increased rates of motor vehicle (1.96; 1.23 to 3.14), and assault (1.90, 1.16 to 3.15), injuries were identified among men who were current users; an increased rate of assault was suggestive in women (2.21; 0.92 to 5.19). CONCLUSIONS: Though the results must be viewed cautiously, they suggest that marijuana use may be independently associated with increased risk of hospitalized injury. Further study of the physiological and behavioral mechanisms is warranted [corrected].