The present study investigated whether IVUS could serve as a reliable reference in validating MDCT characterization of coronary plaque against a histological gold standard. Twenty-one specimens were postmortem human coronary arteries. Coronary cross-sections were imaged by 40 MHz IVUS and by 64-slice MDCT and characterized histologically as presenting calcified, fibrous or lipid-rich plaques. Plaque composition was analyzed visually and intra-plaque MDCT attenuation was measured in Hounsfield Units (HU). 83 atherosclerotic plaques were identified. IVUS failed to characterize calcified plaque accurately, with a positive predictive value (ppv) of 75% versus 100% for MDCT. Lipid-rich plaque was even less accurately characterized, with ppv of 60 and 68% for IVUS and MDCT respectively. Mean MDCT attenuation was 966 +/- 473 HU for calcified plaque, 83 +/- 35 HU for fibrous plaque and 70.92 HU +/- 41 HU for lipid-rich plaque. No significant difference in mean MDCT attenuation was found between fibrous and lipid-rich plaques (P = 0.276). In vivo validation of MDCT against an IVUS reference thus appears to be an unsuitable and unreliable approach: 40 MHz IVUS suffers from acoustic ambiguities in plaque characterization, and 64-slice MDCT fails to analyze plaque morphology and components accurately.