Previous studies show that stress perfusion imaging provides independent and incremental information in risk assessment. These studies, however, did not include quantitative methods. This study examined the predictors of prognosis using quantitative data from treadmill exercise, coronary angiography, and exercise SPECT thallium-201 imaging in medically treated patients with stable symptoms. There were 732 medically treated patients (excluded were patients with recent revascularization or acute ischemic syndromes) who had coronary angiography and exercise SPECT thallium imaging. There were 69 events (cardiac death or nonfatal MI) at a mean follow-up time of 41 ± 22 months. The prognostic values of the clinical characteristics, treadmill exercise score, Gensini score, and SPECT thallium score were examined by Cox survival analysis. Univariate analysis showed that the chi- square of SPECT (χ2 = 15, p = 0.0001) was higher than that of the Gensini score (χ2 = 5, p = 0.03) and the treadmill exercise score (χ2 = 1, p = NS). Actuarial life-table analysis showed that SPECT provided the best separation between groups at high and low risk. The event rates were 13% versus 4% (p = 0.001) by SPECT; 11% versus 7% (p = 0.05) by Gensini score, and 10% versus 7% (p = NS) by treadmill exercise score in groups at high and low risk, respectively. Further, SPECT contained most of the prognostic information provided by coronary angiography. The area under the ROC curve was significantly larger for SPECT than for the Gensini and treadmill exercise scores (p = 0.05). Thus in medically treated patients with stable symptoms, quantitative assessment of exercise perfusion defect size by SPECT is the most important predictor of prognosis; it is significantly more important than the Gensini score (which represents the extent and severity of coronary artery disease) and treadmill exercise score.