Two phases of a back injury prevention program were studied using 2035 accident reports filed between 1979 and 1984. Phase 1, Personnel Program, was designed to decrease the duration of wage-loss claims by increasing the effectiveness of the existing procedures used to process these claims. This program significantly lowered the proportion of high-hour claims (P less than .05) and significantly reversed a trend of increasing accident rates (P less than .05). Phase 2, Back Program, was designed to lower the incidence of back injuries through a feedback-oriented educational program. The Back Program itself could not demonstrate a significant reduction in back injuries primarily due to the powerful and confounding effect of the Personnel Program. The combination of the Personnel Program and the Back Program significantly lowered back injuries for nurses when compared with a similar group of injuries that occurred at geriatric hospitals (P less than .001). The large effect of the Personnel Program and the small effect of the Back Program have design implications for any injury prevention program.