Exercise-induced changes in Q, R, and S wave amplitudes have been reported to detect coronary artery disease but with low specificity, low sensitivity, or both; it was hypothesized that their incorporation into a composite index (Athens QRS score) might improve specificity and sensitivity. For this purpose 246 patients were analyzed retrospectively and 160 prospectively. All patients underwent maximal exercise testing with a standard Bruce protocol and coronary arteriography as part of the diagnostic evaluation for possible or definite coronary artery disease. The Athens QRS score was decreased as the number of obstructed coronary arteries increased (normal coronary arteries = 7.85 +/- 5.23 mm, one-vessel disease = 5.2 +/- 5.3 mm, two-vessel disease = -0.85 +/- 5.4 mm, three-vessel disease = -3.5 +/- 5.8 mm; p less than 0.0001); the score was unrelated to exercise-induced ST segment depression, and negative (less than 0) scores were always associated with coronary artery disease. An Athens QRS score of 5 mm predicted coronary artery disease with sensitivity ranging from 75% to 86% and a specificity ranging from 73% to 79%, values higher than those of the Q wave (75% and 50%, respectively), R wave (65% and 55%), and S wave (70% and 10%) and of the ST segment depression (62% and 70%). It is concluded that exercise-induced changes in the QRS complex provide a useful index not only for the diagnosis but also for the assessment of severity of coronary artery disease.