Clinical and experimental data have documented the ability of nontransmural myocardial infarction to produce abnormal Q waves on both the epicardial and body surfaces. We undertook this study to define the anatomic determinants of such Q wave development. Thirty dogs were studied before and after occlusion-reperfusion (26 dogs) or latex embolization (four dogs) of the left circumflex coronary artery. Occlusion was maintained for 60 to 240 min before reperfusion to produce nontransmural lesions of various sizes. Electrocardiographic data were registered from 84 torso electrodes by body surface mapping techniques before and 1 week after infarction. Infarct size was quantitated by computer analysis of heart slices stained with triphenyl tetrazolium chloride. Six dogs did not develop infarction. Of the remaining 24, 10 did and 14 did not develop significant changes in body surface Q wave duration and width. The incidence of Q wave changes was not different in dogs with nontransmural and those with transmural lesions. Infarct size (expressed as a percentage of the left ventricle infarcted), the percentage of endocardium subjacent to infarction, the average depth of necrosis, the percent of the four outer fifths of the ventricular wall infarcted, and the duration of occlusion were significantly (p less than .05) greater in dogs with than in those without Q wave changes. Logistic regression modeling demonstrated that only two anatomic parameters--percentage of left ventricle infarcted and average lesion depth--significantly and independently predicted Q wave development. A model including only these two variables accurately classified all 24 cases.(ABSTRACT TRUNCATED AT 250 WORDS)