There is significant histopathological and clinical evidence that near-infrared auto-fluorescence (NIRAF) complements optical coherence tomography (OCT) for detecting high-risk coronary plaque. Here, we determined the accuracy of an OCT-NIRAF imaging system and catheter for detecting NIRAF in human coronary lesions. OCT-NIRAF pullback imaging was performed on human cadaver coronary arteries (n=33 from 14 patients) during PBS perfusion via a fully integrated OCT-NIRAF imaging system and catheter (NIRAF ex. 633 nm, 1 mW power; em. 660-740nm). Confocal NIRAF images were acquired from corresponding unstained formalin-fixed paraffin-embedded sections (Olympus FLUOVIEW FV1000; ex. 635 nm; em. 655-755nm). OCT-NIRAF and confocal NIRAF images were registered using known pullback speed, anatomical landmarks, and fiducial features (e.g., calcification), and spatially overlapped by affine transformation of the confocal NIRAF images. Each image was split into 8, 45º-sectors, emanating from the catheter location. Each 45º-sector was determined to be positive if >5% of the intima contained confocal NIRAF, and if >5% of 45º-arc (2.25º) of the catheter-based NIRAF signal was above the system’s detection limit. A total of 1896 45º-sectors from 291 distinct coronary locations were analyzed using confocal NIRAF as the gold standard. Considering superficial confocal NIRAF foci within 0.5 mm from the luminal surface, sensitivity and specificity were 90.0% (95%CI: 69.8-100.0%) and 90.2% (95%CI: 88.8-91.7%), respectively. Within 0.5 mm to 1.0 mm depth from the luminal surface, the sensitivity was 36.4% (95%CI: 15.0-57.8%) and specificity was 90.1% (95%CI: 88.6-91.5%). These results indicate that the OCT-NIRAF system/catheter’s ability to detect NIRAF is depth dependent and accurate in plaque regions (within 0.5 mm from the luminal surface) that are most responsible for precipitating coronary events.