A prospective one-year follow-up study was carried out among 189 civil servants of a municipal social service department. Its aim was to investigate whether psychosocial factors at work predict back pain prevalence at follow-up. The workers were questioned twice about personal characteristics, psychosocial factors at work, physical workload, back pain experience, and general health and well-being. Back pain was assessed as twelve month prevalence and pain intensity on a visual analogue scale (VAS). In a univariate analysis the variables: high job demands, confinement to the workplace, depression, psychological complaints, and general health complaints, were positively related to back pain prevalence at follow-up. In a multivariate analysis, however, none of the odds ratios for psychosocial stressors differed significantly from one another. Among the back pain variables, pain intensity on the VAS was the best predictor of back pain at follow-up. In a final multivariate logistic regression model none of the above-mentioned variables contributed significantly, except for initial back pain. Initial back pain with low intensity and high intensity significantly predicted the prevalence of back pain at follow-up with odds ratios of 3.0 (90% CI: 1.5-6.0) and 10.3 (90% CI: 4.1-25.5), respectively. In conclusion, the present study does not provide clear evidence that psychosocial factors at work predict back pain.