Recent studies have demonstrated that the number of times a hospital or surgeon performs certain procedures annually has an inverse relationship with in-hospital mortality rates for patients undergoing the procedures. This study uses an improved measure of physician volume to test the combined relationship of hospital and physician volume with in-hospital mortality rates and to explore the existence of threshold volumes that optimally discriminate high- and low-volume providers. Five procedure groups have significant volume-mortality relationships. For total cholecystectomies, hospital volume is the more significant volume measure, but physician volume is marginally related to mortality rate. For coronary artery bypass surgeries, resection of abdominal aortic aneurysms, partial gastrectomies, and colectomies, physician volume is more significant than hospital volume, but hospital volume is marginally significant. Annual hospital volume thresholds for these data appear to exist at approximately 5 procedures for partial gastrectomies, 40 procedures for colectomies, and 170 procedures for total cholecystectomies.