The role model displayed by clinician-teachers influences learning experiences but learners may face various reasoning styles. Our goal was to describe common strategies in clinical data collection displayed by experienced clinician-teachers in internal medicine. We studied six internists heavily involved in teaching while they were working up the same seven cases portrayed by a standardized patient. Each encounter was audio-recorded and replayed to allow the subjects commenting on the purpose and diagnostic hypotheses considered for each piece of information collected. Information and hypotheses elicited by all physicians were considered key items. Although the subjects reached the same final diagnoses, they differed on several characteristics of their data collection process. They also displayed common behaviours, such as: early acquisition of key data (half of them acquired within the first 19 questions asked) through clarification of the patients' complaints and focused data collection; early generation of the final diagnosis (within the first 10 questions asked) and use of diagnostic hypotheses to frame data collection; and summarization of the information at hand during the encounter (at least twice). Whether making teachers explicitly conscious about their own reasoning processes may help them better model and explain their diagnostic approach to specific cases should be assessed in follow-up studies.