Background: The admissions interview still remains the most common approach used to describe candidates' noncognitive attributes for medical school. Aim: In this prospective study, we have investigated the predictive validity of a semi-structured interview for admissions to medical school based on medical judgment vignettes: (1) ethical decision-making (moral), (2) relationships with patients and their families (altruistic), and (3) roles and responsibilities in professional relationships (dutiful). Method: A group of 26 medical students from the Class of 2007 participated in the interview process and provided their subsequent performance results from clerkship 3 years later. Results: Inter-rater reliability of the scored interviews was high ( 0.96). Our results provided evidence for both convergent and divergent predictive validity. Medical judgment vignettes scores correlated significantly with seven mandatory clerkship rotation in-training evaluation reports (r 0.39, p < 0.05; to r 0.55, p < 0.01). Conclusion: This semi-structured interview based on clearly defined and scored medical judgment vignettes that focus on the assessment of medical students' noncognitive attributes is promising for student's selection into medical school. The high reliability and evidence of predictive validity of clinical performance over a 3-year period suggests a workable approach to the assessment of 'compelling personal characteristics' beyond merely cognitive variables.