In order to define the coronary lesions we prospectively performed digital coronary angiographies in 61 patients. The degree of stenosis was measured in 100 lesions by quantitative analysis using densitometric and geometric methods. Two groups of lesions were found by comparing these two methods: Group A, 47 lesions with a poor correspondence in the degree of stenosis between densitometric and geometric analysis (p greater than 0.01; and group B, 53 lesions with a good correspondence. Both groups were correlated with plaque characteristics (unstable or stable), following angiographic criteria. The mean degree of stenosis in all lesions, for densitometric and geometric analysis was 50.04 +/- 21.1% and x 60.66 +/- 22.1% (p less than 0.01), respectively. Unstable plaque was more frequent in group A (80.9%) than in B (17.9%) (p less than 0.0001), and stable plaque was more frequent in B (81.1%) than in A (19.1%) (p less than 0.0001). Less degree of stenosis between A (41.5 +/- 13.2) and B (61.3 +/- 16.05) was found by densitometric analysis (p less than 0.0001) but it was similar by geometric method (60.02 +/- 11.3 in A and 58.6 +/- 14.4 in B) so that the degree of stenosis in unstable plaque was lower by densitometric method. We conclude that densitometric analysis showed poor correlation with geometric analysis in unstable plaques; the difference could be due to the soft component expression of the unstable plaque.