A correlation between stress at work and psychosomatic complaints is often given a direct causal interpretation. This study examines four alternative explanations: third variables that produce spurious correlations between subjective stress and psychosomatic complaints, third variables that produce spurious correlations between objective stress and psychosomatic complaints, and the reverse causation hypothesis that psychosomatic complaints lead to more stress at work. A final set is not an actual alternative explanation but maintains that it holds only for special groups, for example those who overreport stress at work. Three studies of German, male, blue-collar workers-two cross-sectional ones (N = 206 and N = 841) and one longitudinal one (N = 90)-measured both physical and psychological stress at work on three levels: subjective response, observers' ratings, and a median rating of people doing the same work. The last two are regarded as more objective than the first. The results show that the correlation between physical stress and psychosomatic complaints can be explained by some third variable, whereas the correlation between psychological stress and psychosomatic complaints cannot be explained by any of the alternative hypotheses. Objective conditions of work stress influence the development of subjective stress and psychosomatic complaints. © 1985 American Psychological Association.