Early caries detection is essential for the implementation of preventive, therapeutic and intervention strategies within general dental practice. Objective: The aim of this study was to compare the in vitro performance of the International Caries Detection and Assessment System (ICDAS), digital photographs scored with ICDAS (ICDAS photographs), fibre-optic transillumination (FOTI), optical coherence tomography (OCT), SoproLife® camera and two implementations of quantitative light-induced fluorescence a commercial (QLF-Inspektor Research systems) and a custom (QLF-Custom) system, to detect early and intermediate occlusal lesions. Methods: One hundred and twelve permanent extracted teeth were selected and assessed with each detection method. Histological validation was used as a gold standard. The detection methods were compared by means of sensitivity, specificity, areas under receiver operating characteristic (AUROC) curves for enamel and dentine levels and with the Spearman's rank correlation coefficient against histology. Results: For any enamel or dentine caries detection, the AUROC curves ranged from 0.86 (OCT) to 0.98 (ICDAS and ICDAS photographs, SoproLife® camera) and at the dentine level from 0.83 (OCT) to 0.96 for FOTI. The correlations with histology ranged between 0.65 (OCT) and 0.88 (ICDAS and FOTI). Under in vitro conditions, the assessed detection methods showed excellent intra-examiner reproducibility. All the methods were strongly correlated with histology (p <0.01) except OCT which showed a moderate correlation (0.65). Conclusion: Even though all methods present similar performance in detecting occlusal caries lesions, visual inspection seems to be sufficient to be used in clinical practice for detection and assessment of lesion depth. Other methods may be useful in monitoring caries lesion behaviour. © 2012 Elsevier Ltd. All rights reserved.