This study evaluated whether risk prediction data captured by nurse case managers (NCMs) could predict prolonged disability from occupational low back pain. Cases of uncomplicated occupational low back pain referred to NCMs, with dates of injury between January 1, 1997 and June 30, 1998, were selected (n = 986). Predictive information was obtained from the NCM screening checklist of potential risk factors and other administrative claims data. Data were collected on 23 variables thought to predict work status at 6 months postinjury. In a multivariate logistic regression model, being out of work was significantly associated with older age, language barriers, earlier referral to NCM, and neutral or negative attorney attitude toward return to work. The overall predictive power of the model was low (area under curve = 0.6). Although NCM risk factor data collection is feasible and practical in insurance settings, future studies should explore additional variables and refined data collection methods in order to achieve a more accurate prediction.