A new treadmill exercise protocol, modified from the standard Bruce method, was designed to improve ST segment/heart rate slope accuracy and applicability by reducing heart rate increments between exercise stages. In 150 patients exercised according to the new protocol and in 150 patients exercised according to the Bruce protocol, similar exercise tolerance and similar overall heart rate, systolic blood pressure and double product responses to exercise were observed. The mean increment in heart rate between exercise stages of the new protocol was ten beats/minute, which was significantly lower than the 27 beats/minute/stage found with the Bruce protocol. The accuracy of computer-measured ST segment depression was validated by comparison with physician measurement in a separate subgroup of patients with angina pectoris, and serial testing demonstrated stronger interest reproducibility for the ST segment/heart rate slope than for either measured ST segment depression, peak heart rate achieved or duration of exercise.