This study was performed to assess the strengths and weaknesses of optical coherence tomography (OCT) intravascular imaging in identifying plaque morphology. Seventy-nine postmortem human coronary arterial sections classified as fibrous-cap atheromas, calcific plaques, fibrous plaques, and complicated lesions were studied. OCT was able to identify 45% of fibrous-cap atheromas (kappa=0.27, p<0.01), 68% of fibrocalcific plaques (kappa=0.40, p<0.001), 83% of fibrous plaques (kappa=0.37, p<0.001), and 100% of complicated lesions (all thrombi; kappa=1, p<0.001). Misinterpretation was caused mainly by the low OCT signal penetration, which could not detect lipid pools or calcium behind thick fibrous caps, and by an inability to distinguish calcium deposits from lipid pools or the opposite. Lesions with thick (>150 microm) caps were histologically identified as 25 thick fibrous-cap atheromas, 8 fibrocalcific plaques, and 5 fibrous plaques; all these lesions were relatively "stable." In contrast, lesions with fibrous caps<150 microm were either vulnerable or stable lesions (11 thin-fibrous-cap atheromas and 11 fibrocalcific plaques). In conclusion, although OCT images may give an indication of the overall composition of large homogenous signal-poor regions, such as lipids or calcified areas, they could be unreliable in differentiating areas with heterogenous compositions. OCT may easily recognize relatively stable lesions.