This study examined the ability of the treadmill exercise score (TES) in determining the presence and extent of coronary artery disease (CAD). The score was derived from the integrated area of ST segment depression and ST slope in two leads (V5 and a VF), corrected for R wave amplitude, exercise time, and percent of maximum predicted heart rate. The ST segment depression was measured at 80 msec after the J point. There were 34 patients with no significant CAD, 38 patients with one-vessel CAD (≥ 50% diameter stenosis), and 58 patients with multivessel CAD. The TES showed a considerable scatter in patients with and without CAD. A receiver operating characteristic curve showed different levels of sensitivity and specificity, depending on the cut-point. The TES was similar to ST segment depression in detecting CAD (predictive accuracy, 77% vs 78%, p = NS). A markedly abnormal score (<-1.0) was seen in 41 patients, of whom 32 (78%) had multivessel CAD. On the other hand, a score > 0 was seen in 49 patients, of whom 40 (82%) had no or one-vessel CAD. In 40 patients with TES between -1.0 and 0, 17 (43%) had multivessel CAD and 23 (57%) had no or one-vessel CAD. In 51 patients with nondiagnostic ST changes, the TES correctly classified the extent of CAD in 20 patients (40%). Thus, the TES has a similar accuracy to the ST segment depression criteria in detecting CAD. The extent of CAD can, however, be ascertained in 80% of the patients with very high or very low TES. © 1987.