During the last two decades two potent groups of predictors for work-related musculoskeletal problems have been identified: physical work load and poor psychosocial working conditions. However, little is known about their combined effects. In this study the buffering effect of control at work with respect to the negative effects of psychological demands and physical work load on musculoskeletal problems is examined. All study variables were sampled from 431 people working in geriatric nursing homes in Germany by means of questionnaires. Main effects and interactions were tested with multiple regression analysis. The results showed that control buffered the effects of high psychological demands, but not of high physical work load; the buffering effect of control was observed only when physical work load was low. The combined effects of demands and physical work load were over-additive. Thus, the power of different predictors for musculoskeletal symptoms depended on the level of other predictors. This suggests that the efficiency of certain strategies for the prevention of musculoskeletal problems is likely to depend on the level of risk factors, with different strategies being appropriate for different levels of risk.