This paper gives a preliminary report on the "critical incidents" technique used for evaluating an experimental program for treating indigent patients and training medical students at the General Medical Clinic of the Denver General Hospital. This program has the experimental group of medical students treat the same patients throughout the course of their treatments in an attempt to restore some of the virtues of the old-time family physician and to provide comprehensive and continuous medical care. The critical incidents technique, developed by John C. Flanagan of the American Institute for Research, in this application, asked a select panel of experienced doctors to describe from their own experiences "critical incidents" where either good or bad medicine was practiced. It was limited by the conditions of the program which couldn't include all circumstances that the students would encounter as doctors, so only 24 practicing physicians were interviewed and only 223 usable incidents were collected and classified. The initial classification is presented here under the general headings doctor-patient relationships, relationships with other members of the medical team, community health problems and resources, relation of medicine to government, personal behavior, and basic knowledge and technical competence.