Exercise electrocardiographic stress testing was performed in 81% of the 780 patients randomized in the Coronary Artery Surgery Study at entry. The cumulative survival at the end of 7 year follow-up was 90% for those assigned to surgical treatment and 88% for those assigned to medical therapy (p = NS). These survival rates did not differ significantly from either those of the entire randomized cohort or those of the 149 patients who did not have a qualifying exercise test at baseline. No differences in important baseline characteristics existed between those who were exercised and not exercised at entry. Stratification of patients according to the degree of ST segment depression (less than 1 mm, greater than or equal to 1 mm, greater than 2 mm) and final exercise stage achieved during a Bruce protocol treadmill test (final stage less than or equal to 1, stage 2 and greater than or equal to stage 3) failed to show any significant differences in 7 year survival rates between medically and surgically assigned patients. Additionally no differences in survival were noted within either the medical or surgical groups regardless of the degree of ST segment depression or the final stage achieved. The presence of exercise-induced angina, however, identified patients who had a survival advantage if assigned to surgical therapy, with a 7 year survival rate of 94% compared with 87% for medically assigned patients (p = .007).(ABSTRACT TRUNCATED AT 250 WORDS)