The purpose of this study was to provide validity and feasibility evidence in measuring professionalism using the Professionalism Mini-Evaluation Exercise (P-MEX) scores as part of a residency admissions process. In 2012 and 2013, three standardized-patient-based P-MEX encounters were administered to applicants invited for an interview at the University of Geneva Pediatrics Residency Program. Validity evidence was gathered for P-MEX content (item analysis); response process (qualitative feedback); internal structure (inter-rater reliability with intraclass correlation and Generalizability); relations to other variables (correlations); and consequences (logistic regression to predict admission). To improve reliability, Kane's formula was used to create an applicant composite score using P-MEX, structured letter of recommendation (SLR), and structured interview (SI) scores. Applicant rank lists using composite scores versus faculty global ratings were compared using the Wilcoxon signed-rank test. Seventy applicants were assessed. Moderate associations were found between pairwise correlations of P-MEX scores and SLR (r = 0.25, P = 0.036), SI (r = 0.34, P = 0.004), and global ratings (r = 0.48, P < 0.001). Generalizability of the P-MEX using three cases was moderate (G-coefficient = 0.45). P-MEX scores had the greatest correlation with acceptance (r = 0.56, P < 0.001), were the strongest predictor of acceptance (OR 4.37, P < 0.001), and increased pseudo R-squared by 0.20 points. Including P-MEX scores increased composite score reliability from 0.51 to 0.74. Rank lists of applicants using composite score versus global rating differed significantly (z = 5.41, P < 0.001). Validity evidence supports the use of P-MEX scores to improve the reliability of the residency admissions process by improving applicant composite score reliability.