Percutaneous coronary intervention is performed by operators with differing experience, technique, and case mix. It is unknown if operator practice patterns impact patient outcomes. We sought to determine if a cluster algorithm can identify distinct profiles of percutaneous coronary intervention operators and if these profiles are associated with patient outcomes. Operators performing at least 25 annual procedures between 2014 and 2018 were clustered using an agglomerative hierarchical clustering algorithm. Risk-adjusted in-hospital mortality was compared between clusters. We identified 4 practice profiles among 7706 operators performing 2 937 419 procedures. Cluster 1 (n=3345) demonstrated case mix and practice patterns similar to the national median. Cluster 2 (n=1993) treated patients with lower clinical acuity and were less likely to use intracoronary diagnostics, atherectomy, and radial access. Cluster 3 (n=1513) had the lowest case volume, were more likely to work at rural hospitals, and cared for a higher proportion of patients with ST-segment-elevation myocardial infarction and cardiogenic shock. Cluster 4 (n=855) had the highest case volume, were most likely to treat patients with high anatomic complexity and use atherectomy, intracoronary diagnostics, and mechanical support. Compared with cluster 1, adjusted in-hospital mortality was similar for cluster 2 (estimated difference, -0.03 [95% CI, -0.10 to 0.04]), higher for cluster 3 (0.14 [0.07-0.22]), and lower for cluster 4 (-0.15 [-0.24 to -0.06]). Distinct percutaneous coronary intervention operator profiles are differentially associated with patient outcomes. A phenotypic approach to physician assessment may provide actionable feedback for quality improvement.