The supplementary value of varying degrees of upsloping ST-segment depression observed during treadmill exercise testing to the accuracy of the exercise ST-segment response for detection of ischemia was determined by employing a reversible thallium-201 (201Tl) defect as the criteria for ischemia. A group of 199 consecutive patients (168 men) with > or = 1 reversible 201Tl defects on quantitative planar perfusion imaging, and a normal group of 366 patients with normal 201Tl scans who achieved > or = 85% of age-predicted maximum heart rate were studied. Upsloping ST-segment depression was subcategorized for > or = 1.0, > or = 1.5, and > or = 2.0 mm of sustained ST-segment depression below baseline at 0.08 seconds after the J-point. If only > or = 1.0 mm of horizontal or downsloping ST-segment depression was designated as abnormal and all upsloping responses as normal, the sensitivity, specificity, positive predictive value, negative predictive value, and accuracy of the exercise electrocardiogram were 49% (98 of 199), 84% (281 of 336), 64% (98 of 153), 74% (281 of 382), and 71% (379 of 535), respectively. If in addition to > 1.0 mm of horizontal or downsloping ST-segment depression, > or = 2.0 mm of upsloping ST-segment depression is considered abnormal, these values were 52% (104 of 199), 81% (272 of 336), 62% (104 of 168), 74% (272 of 367), and 70% (376 of 535), respectively. If > or = 1.5 mm of upsloping ST-segment depression is considered an abnormal response, these values were 59% (117 of 199), 74% (248 of 336), 57% (117 of 205), 75% (248 of 330), and 68% (365 of 535), respectively. Finally, if > or = 1.0 mm of upsloping ST-segment depression is considered abnormal, these values were 71% (142 of 199), 56% (187 of 336), 49% (142 of 291), 77% (187 of 244), and 61% (329 of 535), respectively. Thus, if upsloping ST-segment depression is added to the criteria for a positive exercise test result, sensitivity for detection of ischemia is increased but at the expense of a fall in specificity and a significant decrease in the positive predictive value of the test.