The authors designed the present study to examine the association between individuals' scores on the Jefferson Scale of Physician Empathy (JSPE; M. Hojat, J. S. Gonnella, S. Mangione, T. J. Nasca, & M. Magee, 2003; M. Hojat, J. S. Gonnella, T. J. Nasca, S. Mangione, M. Vergare, & M. Magee, 2002; M. Hojat, S. Mangione, T. J. Nasca, M. J. M. Cohen, J. S. Gonnella, J. B. Erdmann, J. J. Veloski, & M. Magee, 2001), a self-report empathy scale, during medical school and ratings of their empathic behavior made by directors of their residency training programs 3 years later. Participants were 106 physicians. The authors examined the relationships between scores on the JSPE (with 20 Likert-type items) at the beginning of the students' 3rd year of medical school and ratings of their empathic behavior made by directors of their residency training programs. Top scorers on the JSPE in medical school, compared to Bottom scorers, obtained a significantly higher average rating of empathic behavior in residency 3 years later (p < .05, effect size = 0.50). The findings support the long-term predictive validity of the self-report empathy scale, JSPE, despite different methods of evaluations (self-report and supervisors' ratings) and despite a time interval between evaluations (3 years). Because empathy is relevant to prosocial and helping behavior, it is important for investigators to further enhance our understanding of its correlates and outcomes among health professionals.