The aim of this in vivo study was to validate the DIAGNOdent laser fluorescence method and the International Caries Detection and Assessment System (ICDAS) for caries classification against the gold standard, histological examination, for detecting occlusal caries on permanent molars; the thresholds used were the outer enamel (D1), the inner enamel and outer third of the dentine (D2), and the inner two-thirds of the dentine (D3). Patients with non-impacted third molars (n = 43) were recruited from a university clinic. A trained examiner performed the examinations. After the teeth were extracted, the histological criterion was used to determine the severity of the lesions. Intra-examiner agreement (weighted kappa) for ICDAS was 0.60 and reliability (intraclass correlation coefficient) for DIAGNOdent was 0.968. The correlations with the histological reference were weak for DIAGNOdent (rs = 0.369) and moderate for ICDAS (rs = 0.515). The areas under the receiver-operating characteristics (ROC) curve at D1, D2, and D3 were 0.60, 0.69, and 0.91, respectively, for ICDAS and 0.55, 0.65, and 0.92, respectively, for DIAGNOdent. ICDAS and DIAGNOdent proved to be reproducible methods with similar performance in the detection of occlusal carious lesions in dentine. The ability of DIAGNOdent to detect initial enamel lesions was higher than that of ICDAS, but with low specificity. The usefulness of DIAGNOdent as an adjunct method for assessment of initial occlusal caries in permanent molars is questionable.