The hypothesis to be tested was whether characteristics of work situations affect the physiology of people when the effects of individual differences in medical/behavioral factors are minimized. Four prisons with different objective work conditions were chosen. The psychosocial work situation was recorded in a self-administered questionnaire. Morning plasma cortisol, blood pressure, and urinary excretion of catecholamines were measured. Analyses of variance showed that the four prisons differed significantly as to mean levels of decision latitude and skill discretion. Analyses of variance also showed significant differences between the prisons with regard to mean levels of noradrenaline at night during leisure time and a tendency towards differences in work time. The rank order of mean levels of noradrenaline in the four prisons was the same as the inverted rank order of mean levels in self-reported work conditions. Thus it seems that when collective data are used and when adjusted for other variables which might have an association with the dependent variables, differences in self-reported work conditions are reflected in the urinary excretion of noradrenaline. Of the three physiological variables used as dependent variables, urinary excretion of noradrenaline reflected arousal as an effect of work conditions more accurately than the other variables.