OBJECTIVE: Inexpensive and standardized methods to deliver medical education to primary care physicians (PCPs) are desirable. Our objective was to assess the impact of an individualized simulated learning intervention on diabetes care provided by PCPs. RESEARCH DESIGN AND METHODS: Eleven clinics with 41 consenting PCPs in a Minnesota medical group were randomized to receive or not receive the learning intervention. Each intervention PCP was assigned 12 simulated type 2 diabetes cases that took about 15 min each to complete. Cases were designed to remedy specific physician deficits found in their electronic medical record observed practice patterns. General linear mixed models that accommodated the cluster randomized study design were used to assess patient-level change from preintervention to 12-month postintervention of A1C, blood pressure, and LDL cholesterol. The relationship between the study arm and the total of intervention and patient health care costs was also analyzed. RESULTS: Intervention clinic patients with baseline A1C >or=7% significantly improved glycemic control at the last postintervention A1C measurement, intervention effect of -0.19% mean A1C (P = 0.034) and +6.7% in A1C <7% goal achievement (P = 0.0099). Costs trended lower, with the cost per patient -$71 (SE = 142, P = 0.63) relative to nonintervention clinic patients. The intervention did not significantly improve blood pressure or LDL control. Models adjusting for age, sex, and comorbidity showed similar results. PCPs reported high satisfaction. CONCLUSIONS: A brief individualized case-based simulated learning intervention for PCPs led to modest but significant glucose control improvement in adults with type 2 diabetes without increasing costs.