Plaques identified by Coronary CT angiography (CCTA) are important in clinical diagnosis and primary prevention. High-risk plaque features by CCTA have been extensively validated using optical coherence tomography (OCT). However, since their general diagnostic performance and limitations have not been fully investigated, we sought to compare CCTA with OCT among consecutive vessel sections. We retrospectively compared 188 consecutive plaques and 84 normal sections in 41 vessels from 40 consecutive patients referred for chest pain evaluation who had both CCTA and OCT with a median time lapse of 1 day. The distance to reference points were used to co-register between the modalities and the diagnostic performance of CCTA was evaluated against OCT. Plaque categories evaluated by CT were calcified, non-calcified and mixed. The diagnostic performance of CCTA was excellent for detecting any plaque identified by OCT with the sensitivity, specificity, negative and positive predictive values and accuracy of 92%, 98%, 99%, 84% and 93%, respectively. The lower than expected negative predictive value was due to failure of detecting sub-millimeter calcified (≤ 0.25 mm) (N = 12) and non-calcified plaques (N = 4). Misclassification of plaque type accounted for majority of false negative findings (25/41, 61%) which was most prevalent among the mixed plaque (19/41, 46%). There was calcification within mixed plaques (N = 5) seen by CCTA but missed by OCT. Our findings suggest that CCTA is excellent at identifying coronary plaques except those sub-millimeter in size which likely represent very early atherosclerosis, although the clinical implication of very mild atherosclerosis is yet to be determined.