OBJECTIVE: The findings of magnetic resonance and x-ray angiography were compared for assessment of coronary artery stenosis in this validation study. BACKGROUND: Magnetic resonance angiography of the coronary arteries has recently been described, but there has been no comparison with x-ray angiography of localisation or assessment of important characteristics of coronary stenosis. METHODS: A breath hold, segmented k-space, 2D gradient echo imaging technique incorporating fat suppression was used in 39 patients (55 coronary stenoses) with known coronary artery disease. RESULTS: Overall, 47 stenoses (85%) were assessed by magnetic resonance (29 of 33 stenoses in the left anterior descending artery, one of one in the left main stem, 14 of 17 in the right coronary artery, and three of four in the left circumflex artery were detected). There was close agreement between magnetic resonance and x-ray angiography for the distance of the stenosis from the arterial origin (magnetic resonance mean (SD) 27 (16) mm versus x-ray angiography 27 (16) mm, P = NS, mean difference -0.2 mm). The distance to 39 stenoses (83%) agreed to within 5 mm, with increased scatter for more distal stenoses. The severity of magnetic resonance signal loss, assessed visually at the site of stenosis, varied significantly according to the percentage diameter stenosis (F = 30, P < 0.0001); stenosis severity with severe signal loss was 89 (7)%, with partial signal was 70 (16)%, and with irregular wall only 37 (11)%, with significant differences among the three groups (P < 0.001). A significant correlation was found between the proportional magnetic resonance signal loss at the stenosis and the percentage diameter stenosis severity (r = -0.67, P < 0.0001). The length of stenosis measured by magnetic resonance (6 (3) mm) was greater than by x-ray angiography (5 (2) mm, P < 0.006, mean difference +1.1 mm). Spearman's rank test showed that there was significant overestimation of stenosis length by magnetic resonance as stenosis severity increased (rs = 0.34, P < 0.02). CONCLUSIONS: Accurate localisation of coronary stenosis and a qualitative assessment of stenosis severity are possible by magnetic resonance, but stenosis length is overestimated as severity increases, probably because of disturbed patterns of flow with turbulence distal to severe stenoses. Reasonable results for the detection of coronary artery stenosis by magnetic resonance were achieved in this highly selected population, but further progress in imaging techniques is necessary before moving towards appreciable clinical application.