The relative impact of a patient's sex and expressivity (expression of a personal problem) on attitudes of physicians toward patients was assessed using case simulations and questionnaires. Eight simulated cases were used that varied by presenting complaint, patient sex, and inclusion or exclusion of a personal problem. Two non-identical cases were read by each of 253 primary-care physicians, yielding 506 questionnaires for analysis. Of the physicians, 25 per cent believed women were likely to make excessive demands on physician time, although only 14 per cent believed this likely of men (p less than 0.01); women's complaints were judged more likely to be influenced by emotional factors (65 per cent versus 51 per cent in men, p less than 0.01), and were identified as psychosomatic more frequently than were men's (21 per cent versus 9 per cent, p less than 0.01). No sex differences were observed for tranquilizer prescriptions. Sex differences persisted when complaint and expressiveness were controlled; however, physicians' reactions to expressivity were strong enough to equalize male-female differences in some items. Although non-expressive women were more likely to receive a psychosomatic diagnosis than non-expressive men (14 per cent versus 2 per cent, p less than 0.01), expressive men and women were almost equally likely to receive psychosomatic diagnoses. Thus, differences in labeling occurred as a function of the patient's sex and expressivity. The effects of these differences on quality of care remain to be determined.