PURPOSE: To evaluate the prognostic value of myocardial perfusion PET in patients with and patients without diabetes mellitus. MATERIALS AND METHODS: The authors performed a retrospective analysis of prospectively acquired data from a multicenter registry cohort of 7061 patients, including 1966 with diabetes mellitus, who underwent clinically indicated rest-stress rubidium 82 (Rb) myocardial perfusion PET. The mean patient age (±standard deviation) was 63.3 years ± 13. Of the 7061 patients, 3348 were women (47.4%), 2296 (32.5%) had known coronary artery disease, and 1895 (26.8%) had previously undergone revascularization. The primary end point was cardiac death ( = 169) assessed at a mean of 2.5 years ± 1.5. The authors used Cox proportional hazards models and risk reclassification measures stratified according to diabetes status. RESULTS: In multivariable models adjusting for established clinical predictors, increasing magnitude of stress myocardial perfusion abnormality was associated with greater risk of cardiac death in patients with diabetes (hazard ratio [HR]: 7.2; 95% confidence interval [CI]: 3.1, 16.8) for severely abnormal myocardium compared with normal myocardium. The addition of stress myocardial perfusion imaging results significantly improved the fit of a clinical model for predicting cardiac death in patients with and patients without diabetes. Myocardial perfusion PET improved risk reclassification for cardiac death in patients with diabetes (category-based net reclassification index: 0.39; 95% CI: 0.15, 0.60, < .001). Among diabetic patients, an abnormal myocardial perfusion PET scan was associated with increased risk of cardiac death (HR: 4.4; 95% CI: 2.0, 9.7) in all important clinical subgroups based on age, sex, obesity, or prior revascularization. CONCLUSION: In a large cohort of patients referred for clinical Rb stress PET, myocardial perfusion imaging results provided incremental risk prediction of cardiac death in patients with and patients without diabetes mellitus.© RSNA, 2019