Forty first year medical students, previously screened for the presence or absence of the Type A behavior pattern, collected saliva samples for four days leading up to the forst major examination of their medical school course. It was observed that the day of the examination produced substantially different patterns of salivary cortisol concentrations for Type A and non-A subjects, depending on their level of performance. The Type A subjects who scored above the median on the examination showed higher concentrations of salivary cortisol than Type A subjects who scored below the median. The reverse of this pattern was found for non-A subjects. Visual analog scale ratings of subjective stress showed no behavior pattern effects and a very low correlation with the salivary cortisol concentrations. The results support the view that the occurrence of physiological arousal associated with the Type A behaviour pattern is intermittent and dependent on situational factors such as expectation of success and actual success. This is consistent with conceptualizing the Type A pattern as a strategy for resource allocation which may be elective, rather then the behavioural manifestation of an automatic physical response to stress.