This study aimed to evaluate a laser fluorescence device (the DIAGNOdent) and a visual classification system (ICDAS-II) for occlusal caries diagnosis. It also aimed to determine whether fluorescence measurements taken at the tooth surface correlate with the fluorescence measurements taken within the body of the lesion. The occlusal surfaces of 100 extracted permanent teeth were examined using ICDAS-II and DIAGNOdent (LF-tooth). Serial sections were made and lesion depth was assessed histologically. DIAGNOdent readings were also taken from the sections (LF-section). There were significant positive strong correlations between ICDAS-II and histology (r(S) = 0.71) and LF-section and histology (r(S) = 0.70), and only moderate correlations between LF-tooth and histology (r(S) = 0.51) and LF-tooth and LF-section (r(S) = 0.60). Diagnostic accuracy for ICDAS-II was generally better than for LF-tooth. While the DIAGNOdent device provides an objective reading for detection and monitoring of carious lesions, using the cut-off ranges previously suggested leads to inferior performance.