A 3-step, 3-segment scintigraphic model was developed to improve the accuracy of dipyridamole-thallium imaging for preoperative cardiac risk assessment and to simplify the prognostic interpretation of the images. The model was developed in a pilot study of 60 patients and validated in a group of 355 patients referred for vascular and major general surgery. Study end points included myocardial infarction and cardiac death. Step 1: The postoperative cardiac event rate was 1.3% in 225 patients with normal anterior, inferio- and posterolateral segment perfusion and without transient left ventricular dipyridamole-induced cavitary dilation. Step 2: The physiologic rationale for step 2 consists of identifying patients who are most likely to have left main, 3-vessel or high-risk 2-vessel coronary artery disease or a significant amount of jeopardized myocardium in the territory of a critical coronary stenosis. Of 29 patients with either reversible defects of all 3 segments, transient cavitary dilation, or at least 1 severe grade 3/3 reversible defect, 52% (15 of 29) sustained a postoperative cardiac event. Step 3: The remaining 101 patients were stratified according to age greater than 70 years (p = 0.01), presence of diabetes (p = 0.0004) and the number of segments displaying reversible defects (1 or 2) with cardiac event rates ranging from 5 to 36%. The 3-step, 3-segment model is a useful alternative to the conventional interpretation of dipyridamole myocardial perfusion images for the purpose of quick and efficient preoperative risk stratification based on the rationale of correlating surgical risk with the amount of potentially ischemic myocardium.(ABSTRACT TRUNCATED AT 250 WORDS)