Recently, fear-avoidance models have been quite influential in understanding the transition from acute to chronic low back pain (LBP). Not only has pain-related fear been found to be associated with disability and increased pain severity, but also treatment focused at reducing pain-related fear has shown to successfully reduce disability levels. In spite of these developments, there is still a lack in well-designed prospective studies examining the role of pain-related fear in acute back pain. The aim of the current study was to prospectively test the assumption that pain-related fear in acute stages successfully predicts future disability. Subjects were primary care acute LBP patients consulting because of a new episode of LBP (<or=3 weeks). They completed questionnaires on background variables, fear-avoidance model variables and LBP outcome (Graded Chronic Pain Scale, GCPS) at baseline, 3, 6, and 12 months follow-up and at the end of the study. Two-hundred and twenty-two acute LBP patients were included, of whom 174 provided full follow-up information (78.4%). A backward ordinal regression analysis showed previous LBP history and pain intensity to be the most important predictors of end of study GCPS. Of the fear-avoidance model variables, only negative affect added to this model. Our results do not really support the longitudinal validity of the fear-avoidance model, but they do feed the discussion on the role of pain-related fear in early stages of LBP.