OBJECTIVE This study provides information on student factors associated with a career choice in family practice. METHODS Information was used from multiple surveys completed by medical students, including the Premedical Questionnaire, the Matriculating Student Questionnaire, and the Graduation Questionnaire, as well as information from residency directors about residents in the Graduate Medical Education Tracking Census. These questionnaires are all a part of the Student and Applicant Information Management System of the Association of American Medical Colleges. Participants were 30,789 students graduating from US medical schools in 1991 and 1992. Comparisons were made between longitudinal student responses on the surveys to four types of outcomes. RESULTS A total of 1,029 (3.3%) students were in the "Maintained" group (students who originally planned to enter family practice and were in a family practice residency at postgraduate year 1; 1,958 (6.4%) were "Gained" (originally chose a specialty other than family practice but entered a family practice residency); 1,950 (6.3%) were "Lost Interest" (originally identified family practice but entered another residency-two thirds of whom selected non-primary care specialties); 21,573 (70.1%) were "Never Interested" (did not express an early interest nor select a family practice residency); and the remainder (13.9%) had incomplete specialty data. Of those originally interested in family practice, 34.5% entered family practice residencies. Only 8.3% of those not originally interested entered family practice residencies. The four groups of students differed on many demographic, attitudinal, and experiential characteristics. Prestige, income, opportunities for research, and faculty status were more important to future specialists, while emphasis on primary care and prevention and practice in smaller communities were more important to the future family physicians. CONCLUSIONS Medical schools could potentially increase the number of students selecting family practice residencies through both admissions policies and medical school experiences. These data provide some specifics on how to recruit students and prevent loss of those originally interested in family practice.