One hundred seventy-one nurses had their back strength evaluated on an isokinetic lifting device and filled out an epidemiologic questionnaire. They were then followed prospectively for 2 years to determine the incidence of job-related low-back injuries. The data were analyzed to determine if the injury incidence correlated with any of the strength or epidemiologic variables collected during the original evaluation. Average peak force measured during the isokinetic lift was 63.8 kg + 13.6 kg at a lift speed of 30.5 cm/sec and 59.1 kg + 14.9 kg at a lift speed of 45.7 cm/sec. Sixteen nurses reported an occurrence of job-related low-back pain or injury during the 2-year prospective period. Discriminate statistical techniques showed that none of the strength or epidemiologic variables correlated with the incidence of pain or injury or explained significant amounts of variance when the variables were regressed on strength or work calculated from the lift force/lift height data. It was concluded that in this high risk population, in which loads are heavy and lifting postures are variable, the use of low-back strength or prior history of pain or injury are poor predictors as to subsequent low-back pain or injury.