To investigate the reproducibility and prognostic significance of an exercise-induced decrease in systolic blood pressure, 47 patients were identified who manifested such a reduction below the pre-exercise standing level in a consecutive series of 436 patients who underwent treadmill exercise testing and cardiac catheterization during a 3 year period. The prevalence of this abnormal finding was 11 percent in the total group but 21 percent in the 124 patients with three vessel or left main coronary artery disease. Patients with an exercise-induced reduction in systolic blood pressure were more likely to be male, have typical angina pectoris with class III or IV functional limitation and to have had a prior myocardial infarction than were patients without this finding (p less than 0.05). Although no complications occurred during the exercise test of these 47 patients, the majority had severe ischemic responses and 14 (30 percent) showed complex repetitive ventricular arrhythmias. Of the 47 patients, 24 (group 1a) received medical treatment and 23 (group 1b) underwent coronary bypass surgery. On repeat exercise testing in 42 patients, a decrease in systolic blood pressure during exercise was consistently present in group 1a (17 of 20) but entirely absent (0 of 22) in group 1b (p less than 0.001). The mean treadmill time, peak heart rate and systolic blood pressure were not significantly different in the initial and on repeat exercise tests in patients in group 1a; however, in patients in group 1b, all of these variables were significantly higher in the repeat test (p less than 0.001). At a mean follow-up time of 37 months, the total cardiac mortality rate was 8 percent (2 of 24) in group 1a and 4 percent (1 of 23) in group 1b. It is concluded that a decrease in systolic blood pressure during exercise testing is highly reproducible and appears to be reversed by coronary bypass surgery.