This study compared thallium stress testing and exercise changes in QRS duration using a computerized 'optic scanner' in three groups. Group 1 consisted of 108 subjects with positive exercise ECG tests by ST-T segment criteria and with proven coronary artery disease. Group 2 included 19 subjects with nondiagnostic exercise ECG ST-T changes and with proven coronary artery disease. Group 3 was formed by 38 healthy controls. Group 1: Mean increase in exercise QRS width of 12.4 +/- 14 ms. Group 3: Mean decrease in exercise QRS width of 4.9 +/- 9.3 ms (p < 0.0001). Group 2: Mean QRS prolongation of 7.8 +/- 9.2 ms, which was significantly different from the controls (p < 0.0001) but not from group 1. When compared to thallium stress testing, exercise QRS prolongation had a sensitivity of 93%, specificity of 71%, relative risk of 5, and positive predictive value of 86%. QRS duration measurement can improve the diagnostic accuracy of the exercise ECG stress test.