This was a two-step study performed to assess the validity of a computer system in the quantification of coronary arterial angiography (CAG). First, the in vitro study involved 10 aluminum tubes in different internal diameters but the same thickness which were pre-filled with 76% urograffin followed by cinefilm taking. Acquisition and digitization of the stop film was undertaken by a videocamera linking to a computer system. The internal diameters and mean gray levels of all tubes were measured to compare with the real diameters and areas. Second, the in vivo study calculated the percentage of stenosis by visual estimation, diameter measurement and gray level measurement by the computer system from 22 vessels in 13 coronary angiograms. The percentage of patency is the ratio of the diameter or gray level from stenotic region to that of normal region in the CAG. The calculated interobserver variability by visual estimation, diameter measurement and Mean gray level measurement were studied. Good correlation was found between measured diameters and true tube diameters (r = 0.99), and between gray level and tube areas (r = 0.94). The diameter measured by computer and the gray level had lower variability (4.0% and 5.7% respectively) than visual estimation (9.8%) did. Thus, the diameter and densitometric gray level measurements might have clinical implications to assess the coronary arterial stenosis from the coronary angiograms.