This study assessed the validity and reproducibility of different combinations of occlusal caries detection methods: visual examination (VE), laser fluorescence (LF) and radiographic examination (RE). Intra- and interexaminer reproducibilities were also assessed. Forty-seven extracted human molars were used and 121 sites, either suspected or not to be carious, were chosen. Occlusal surfaces were examined by 8 volunteers, assigned to three groups according to their level of knowledge and clinical experience on dental practice: group I, undergraduate students; group II, postgraduate students; group III, professors. Three combinations of methods were tested: A: VE + LF, B: VE + RE, C: VE + LF + RE. The examiners scored the sites using ranking scales and chose a final score based on their clinical experience. The gold standard was determined by histological examination of the sites. In general, LF and RE yielded poorer results than the combinations of methods. For combination A, group III showed the highest sensitivity, while group II showed the highest specificity. For combination B, group II showed moderate sensitivity whereas groups I and III exhibited low sensitivities; all groups of examiners reached substantial specificity. For combination C, all groups exhibited moderate sensitivity and substantial specificity. Interexaminer reproducibility ranged from fair to moderate for combinations A and C, while for combination B kappa values indicated moderate interexaminer reproducibility. It may be concluded that individual exams presented inferior performance than the conjunction of them. Combination C (VE + LF + RE) resulted in the best accuracy for all groups. The knowledge background of the examiners influenced their ability to detect caries lesions and affected interexaminer reproducibility.