This laboratory study of 100 occlusal surfaces investigated the reproducibility and accuracy of a visual ranked caries scoring system, an electronic caries scoring system (ECM) using a continuous conductance scale, and a radiographic ranked caries scoring system. Histological examination of the teeth served as a gold standard to validate the ability of each system to assess lesion depth and predict softened, demineralized dentine. After training, 3 examiners carried out each scoring system on two separate occasions. Kappa values for visual, ECM and radiographic ranked scoring systems showed good inter- and intra-examiner reproducibility levels and acceptable limits of agreement for ECM readings. When scoring systems were tabulated against histological scores there was a high correlation between the visual and ECM methods and lesion depth in both enamel and dentine, but radiographic examination could not detect enamel caries. When compared to the histological scoring, the Spearman correlation coefficients for the visual scoring ranged between 0.87 and 0.93, for the ECM between 0.80 and 0.85 and for the radiographic scoring system between 0.76 and 0.78. No tooth scored as visually sound had histological evidence of dentine caries. Soft dentine corresponded to demineralization involving the middle third of the dentine or more which was related to visual cavity formation or an ECM reading above 9 (score 3 or 4). The radiograph was an excellent predictor of soft dentine. In conclusion, the new visual system appears promising, but takes time to learn. The reproducibility and accuracy for the ECM is acceptable while radiographs miss early occlusal lesions.