The purpose of this study was to establish outcome measures for professionalism in medical students and to identify predictors of these outcomes.Retrospective cohort study.A US medical school.All students entering in 1995 and graduating within 5 years.Outcome measures included review board identification of professionalism problems and clerkship evaluations for items pertaining to professionalism. Pre-clinical predictor variables included material from the admissions application, completion of required course evaluations, students' self-reporting of immunisation compliance, students' performance on standardised patient (SP) exercises, and students' self-assessed performance on SP exercises.The outcome measures of clerkship professionalism scores were found to be highly reliable (alpha 0.88-0.96). No data from the admissions material was found to be predictive of professional behaviour in the clinical years. Using multivariate regression, failing to complete required course evaluations (B = 0.23) and failing to report immunisation compliance (B = 0.29) were significant predictors of unprofessional behaviour found by the review board in subsequent years. Immunisation non-compliance predicted low overall clerkship professional evaluation scores (B = - 0.34). Student self-assessment accuracy (SP score minus self-assessed score) (B = 0.03) and immunisation non-compliance (B = 0.54) predicted the internal medicine clerkship professionalism score.This study identifies a set of reliable, context-bound outcome measures in professionalism. Although we searched for predictors of behaviour in the admissions application and other domains commonly felt to be predictive of professionalism, we found significant predictors only in domains where students had had opportunities to demonstrate conscientious behaviour or humility in self-assessment.