In part 1 of a three-part study, 14 novice readers and 6 experienced cardiologists interpreted phantom images of known stenosis severity. No difference between the interpretations of experienced and novice readers was detectable. Visual estimates of "moderately" severe stenosis were 30% higher than actual percent diameter stenosis. In part 2 of the study, visual interpretation of percent diameter stenosis from 212 stenoses on 241 arteriograms was compared with quantitative coronary arteriographic assessment. The visual analysis overestimated disease severity in arteries with greater than or equal to 50% diameter stenosis (except for right coronary lesions) and underestimated severity in all arteries with less than 50% diameter stenosis. Of the 241 arteriograms, 40 had quantitative and visual analysis of all three coronary arteries for assessment of significant disease. In only 62% of the cases did visual and quantitative methods agree on the presence of severe disease; visual estimates diagnosed significantly (p less than 0.05) more three-vessel disease. In part 3 of the study, comparison of percent diameter stenosis by visual estimate with quantitative coronary arteriographic assessment before and after balloon angioplasty of 38 stenoses showed that visual interpretation significantly (p less than 0.001) overestimated initial lesion severity and underestimated stenosis severity after angioplasty.