The clinical use of exercise rehabilitation programs has increased for patients with coronary artery disease. Exercise testing in these programs typically is conducted on a treadmill or cycle ergometer, although many patients' vocations require upper extremity activities and some patients cannot perform lower extremity exercises. To compare the hemodynamic responses and the incidence of angina and ST-segment depression during upper and lower extremity exercise in patients with coronary artery disease, we administered symptom-limited arm ergometer and submaximal or maximal symptom-limited treadmill tests to 95 cardiac rehabilitation patients who had completed an eight-week exercise training program. Treadmill testing resulted in significantly higher heart rates, systolic blood pressures, and double products than arm ergometer testing. The incidence of ST-segment depression was significantly greater with treadmill testing than with arm ergometer testing, but the incidence of angina was not different between tests. Ten patients had ST-segment depression during both arm ergometer and treadmill testing, and the double products at the onset of ST-segment depression were not different. Our data suggest that arm ergometer testing is less likely to result in ST-segment depression than treadmill testing in patients with coronary artery disease, possibly because of the lower hemodynamic responses during arm ergometer testing.