Exercise stress electrocardiography is the recommended method for cardiac evaluation of patients with normal electrocardiograms (ECGs). There are no data to indicate an independent value of myocardial perfusion imaging (MPI) in predicting mortality in these patients. This study assessed the value of exercise stress MPI in predicting mortality in patients with normal baseline ECGs. We studied 319 patients (55 +/- 10 years of age; 180 men) with normal ECGs by exercise stress technetium-99m tetrofosmin MPI. End points during follow-up were cardiac and all-cause mortalities and hard cardiac events. A normal scan was detected in 190 patients (60%). Myocardial perfusion abnormalities were fixed in 59 patients (18%) and reversible in 70 (23%). During a mean follow-up of 7 +/- 1.2 years, 46 patients (14%) died. Death was considered cardiac in 28 patients (9%). Nonfatal myocardial infarction occurred in 12 patients (4%). Annual cardiac death rates were 0.4% in patients with normal perfusion, and 2.7% in patients with reversible defects. Annual total mortality rates were 1.1% in patients with normal perfusion and 3.4% in patients with reversible defects. In a multivariate analysis model, reversible perfusion abnormalities were associated with cardiac death (RR 2.8, 95% confidence interval 1.6 to 5.1) and hard cardiac events (RR 2.7, 95% confidence interval 1.5 to 4.5). Perfusion abnormalities in multivessel distribution were predictive of all-cause mortality (RR 2, 95% confidence interval 1.4 to 3.2). ST-segment depression was not significantly associated with events. In conclusion, stress technetium-99m tetrofosmin MPI provides independent information for predicting cardiac and overall mortalities in patients with normal ECGs. Reversible perfusion abnormalities, but not ischemic electrocardiographic changes, are predictive of outcome in these patients.