An attending tutor program designed to increase faculty-student contact within an Obstetrics and Gynecology clerkship was evaluated. Sessions were observed, written documents were reviewed, and faculty and students were interviewed in order to determine if the program was meeting its goals. Based on the evaluation data, the program was restructured to increase the tutors' responsibility for student evaluation. The evaluation design was analyzed utilizing the criteria of trustworthiness proposed by Guba. Certain issues which may have relevance for investigators using naturalistic approaches in clinical settings were raised. (Author)