The diagnostic methods of third year residents in internal medicine (N=23) and family practice (N=22) were compared with respect to common ambulatory patient problems. Five written simulated patients were presented and the dependent variables were: initial and revised diagnostic hypotheses, physical examination items, and laboratory charges. The two groups considered the same number and type of diagnostic hypotheses. There were large differences in the selection of physical examination items (P less than .001), with the family practice group selecting fewer items. Laboratory charges were significantly greater for the internal medicine group with two patients (P less than .05), and the charges were nearly identical with two patients. A high degree of patient-specific behavior was demonstrated by both groups. These findings have implications for the future training of primary care physicians.