BACKGROUND The incremental prognostic value of rest-redistribution (201)Tl compared with stress and rest perfusion abnormalities has not been defined. RESULTS We identified 458 patients who underwent rest (201)Tl /stress (exercise or adenosine) (99m)Tc sestamibi single-photon emission computed tomography (SPECT) and had late (18 to 24 hours) (201)Tl imaging, were not revascularized within 60 days of SPECT, and were followed up at >1 year. SPECT images were visually analyzed with the use of a 20-segment model on a scale of 0 to 4. Thirty-seven cardiac deaths (CDs) and 17 nonfatal myocardial infarctions occurred. Univariate Cox proportional hazards analysis showed that the presence of a large amount of rest (201)Tl reversibility (rest-late summed difference score [SDS] of >8) was a significant predictor of CD (chi(2) = 5.77, P = 0.02) and CD or myocardial infarction (chi(2) = 5.3, P = 0.02). The CD rate was 9.3% y(-1) in patients with rest-late SDS of >8 compared with 3.6% y(-1) in patients with a mild/moderate amount of rest reversibility (rest-late SDS 3 to 8) and 3.4% y(-1) in patients with no rest reversibility (rest-late SDS <3) (P = 0.029). Kaplan-Meier survival analysis demonstrated significantly lower cumulative survival rates in patients with rest-late SDS of >8 (P = 0.01). Multivariate Cox proportional hazards analysis demonstrated that the presence of a large amount of resting reversibility was an independent and incremental predictor of CD after adjustment for stress and rest perfusion information. Multivariate logistic regression analysis demonstrated that resting reversibility was not an independent predictor of referral to coronary angiography and revascularization. CONCLUSIONS The identification of a large amount of resting (201)Tl reversibility is an independent predictor of CD over stress and rest perfusion abnormalities.