Objectives: To use methods from the field of marketing research to involve students in the redesign of McMaster University's small group, problem-based undergraduate medical education program. Methods: We used themes from a focus group conducted in an electronic decision support lab to compose 14 four-level educational attributes. Undergraduate medical students completed a discrete choice experiment composed of 15 web-administered, partial-profile, conjoint-choice tasks. Results: Latent class analysis revealed two segments with different preferences. Segment 1, (86% of students), preferred a problem-based approach with more small group tutorial sessions led by expert tutors who facilitated the tutorial process without teaching didactically. Segment 2, (14% of students), preferred more large group lectures, explicit learning objectives, expert tutors who taught didactically, and streaming options based on learning preferences. Although Segment 1 preferred smaller tutorial groups, simulations predicted these students would trade increases in tutorial group size for a conceptually integrated program that included tutorial problems based on core curriculum concepts, greater integration of the content of clinical skills training sessions and the tutorial curriculum, and a link between clerkship patient selection and the program's curriculum. A majority of both segments would accept a more conceptually integrated program if the savings associated with increases in tutorial group size was reinvested in web-enhanced tutorial processes and computer-simulated health care problems. Conclusions: Most students preferred a small group, web-supported, problem-based learning approach led by content experts who facilitated group process. Students favored a program in which tutorial group problems, clinical skills training sessions and the patients selected for clerkship activities were more closely linked to core curriculum concepts.