We analyze the effects of relative income expectations, expected malpractice premium cost, and other economic and noneconomic factors on physician specialty choice. The data for this paper are taken from responses of medical students who completed the Association of American Medical Colleges' Medical School Questionnaire and graduated from medical school in 1995. A random utility model is used to guide our thinking; the econometric technique is multinomial logit regression. Selection of a surgical or support specialty is found to be positively income motivated, while the influence of expected relative income is negatively related to the choice of primary-care and medical practices. Concern over malpractice premium cost is negatively related to surgical and positively related to primary-care selection. Other important determinants of choice are planned location of practice, length of residency, type of medical school attended, score on the science problems section of the Medical College Admission Test, predictable working hours and perceived prestige of the specialty. Policies that alter expected relative income, length of residency, desired location of practice, medical school attended, predictable working hours, and prestige of practice, rather than financial aid, may be appropriate for correcting a perceived maldistribution of physicians among specialties.