The clinical value of exercise-induced variations in ST-segment depression and R wave amplitude in consecutive sinus beats was studied in 160 patients who had a positive treadmill exercise test with the Bruce protocol. The patients, all of whom underwent cardiac catheterization, included 100 with coronary artery disease (CAD) (group with true positive test) and 60 with normal coronary arteries (group with false positive test). Minimal or no exercise-induced variations in the magnitude of ST-segment depression despite variations in R wave amplitude were observed in 84 of the 100 patients with CAD and in only 9 of the 60 patients with normal coronary arteries (P < .0001). Significant exercise-induced variations in ST-segment depression were observed in only 16 of 100 patients with CAD and in 51 of 60 patients with normal coronary arteries (P < .0001). The coefficient of variation of R wave amplitude was similar in both groups (no statistical significance), while the coefficient of variation of ST-segment depression was much greater in the patients with normal coronary arteries than in those with CAD (P < .0001). It is concluded that variability of ST-segment depression at peak exercise may discriminate false positive from the true positive exercise tests, improving the diagnostic ability of the method.