We compared three methods of coronary stenosis (S) sizing in a multihospital PTCA restenosis trial: visual analysis by independent observers, single point caliper measurements, and quantitative coronary angiography (QCA). Cine films from 50 patients were submitted from the clinical site to the quantitative angiographic center, where visual analysis and computer quantitation were performed. Regression analysis revealed a correlation coefficient of .857 for caliper vs. QCA (p less than .0001) and .876 for visual observation vs. QCA (p less than .0001). No significant differences were found between any of the 3 methods for pre- or post-PTCA stenosis values. However, QCA yielded smaller PTCA changes in stenosis severity than either caliper or visual observation (p less than .05). Both caliper and visual methods correlated well with QCA in assessing stenosis size pre- and post-PTCA. Trained observers can visually assess lesion severity with accuracy approaching QCA. QCA is likely to be less expensive when compared to visual analysis by three experienced observers and should be the method used for estimating the results of PTCA in clinical trials.