Objective Our purpose was to report the prevalence of abnormal treadmill test responses and their association with mortality in a large consecutive series of patients referred for standard diagnostic exercise tests, with testing performed and reported in a standardized fashion. Background Exercise testing is widely performed, but an analysis of responses has not been presented for a large number of consecutive tests performed on patients referred for diagnosis of cardiac disease. Methods All patients referred for evaluation at 2 university-affiliated Veterans Affairs Medical Centers who underwent exercise treadmill tests for clinical indications between 1987 and 2000 were determined to be dead or alive according to the Social Security Death Index after a mean 5.9-year follow-up. Patients with established heart disease (ie, prior coronary bypass surgery, myocardial infarction, or congestive heart failure) were excluded from analyses. Clinical and exercise test variables were collected prospectively according to standard definitions; testing and data management were performed in a standardized fashion with a computer-assisted protocol. All-cause mortality was used as the end point for follow-up. Standard survival analysis was performed, including Kaplan-Meier curves and a Cox hazard model. Results After the exclusions, 3974 men (mean age 57.5 ± 11 years) had standard diagnostic exercise testing over the study period with a mean of 5.9 (±3.7) years of follow-up (64% of all tested). There were no complications of testing in this clinically referred population, 82% of whom were referred for chest pain, risk factors, or signs and symptoms of ischemic heart disease. Five hundred forty-nine (14%) had a history of typical angina. Indications for testing were in accordance with published guidelines. A total of 545 died, yielding an annual mortality rate of 1.8%. The Cox hazard model chose the following variables in rank order as independently associated with time to death: change in rate pressure product, age greater than 65 years, METs less than 5, and electrocardiographic left ventricular hypertrophy. A score based on these variables classified patients into low-, medium-, and high-risk groups. The high-risk group with a score greater than 3 has a hazard ratio of 4 (95% confidence interval 3.82-4.27) and an annual mortality rate of 4%. Conclusion This comprehensive analysis provides rates of various abnormal responses that can be expected in men referred for diagnostic exercise testing at typical Veterans Administration Medical Centers. Four simple variables combined as a score predict all-cause mortality after clinical decisions for therapy are prescribed. (Am Heart J 2001;142:127-35.)