The hypothesis that perceived psychosocial work situation is associated with pain threshold was tested on a sample of 103 men and women aged 19-65 yr in Stockholm. Half of the studied sample was a random sample of men (N = 26) and women (N = 31), while the remaining subjects were medical secretaries (women, N = 28) and furniture movers (N = 31). Pain thresholds were measured by means of an algometer before, during and after a standardized colour word test. The measurements were made on six different points in the neck and shoulder region. Before psychological stress in the laboratory, perceived psychological demands were significantly associated with pain threshold--the higher the demands the higher the pain threshold. During stress those who reported low decision latitude and high degree of sleep disturbance were shown to have a low pain threshold. The findings are consistent with the hypothesis that subjects with high demand levels have an elevated pain threshold when they are not under excessive psychological stress. During psychological stress, on the other hand, those with low decision latitude are more pain sensitive than others, and this is aggravated in those who also report a high degree of sleep disturbance.