OBJECTIVES: Although payers are increasingly reporting on hospital volume and mortality to rank hospitals, the value of these data is uncertain. There is a need for better quality measures, which can be accomplished by utilizing a simple measure, such as mortality and hospital volume, to create and validate the composite quality score. METHODS: We used the Medicare Analysis Provider and Review files for resection of pancreatic cancer (2003-06) and the empirical Bayes approach to combine mortality rates with information on hospital volume at each hospital to create an index. This index determined which weights observed mortality according to how reliable it is estimated, with the remaining weight placed on hospital volume. We validated our index by a) establishing the extent to which it explained hospital-level variation in risk-adjusted mortality rates, and b) determining how well it predicted future hospital performance. RESULTS: Since the average hospital caseload was only 6, 84% of the weight was placed on volume and 16% on mortality. Composite measure explained the highest percentage of hospital level variation (54%) and predicted the largest differences in future risk-adjusted mortality across hospitals (odd ratio=3.16, p=0.000). CONCLUSIONS: Using national Medicare data for resection of pancreatic cancer, we found that simple composite measure was a strong predictor of subsequent performance for operations, and it was more effective than individual measures. Such measures would be useful to help patients and payers identify high and low quality hospitals for major surgery.