Objectives: Although clinical guidelines recommend that dentists should take two bitewing radiographs in children to detect caries lesions in primary molars, the actual benefit of the adjunct methods is unclear. Therefore, we investigated the performance of visual inspection and additional radiographic and pen-type laser fluorescence (DIAGNOdent pen - LFpen) methods in detecting proximal and occlusal caries lesions of primary molars. Methods: The subjects were randomly chosen from a pool of enrolment forms of children (4 to 12 yrs-old) who had looked for dental treatment. Primary molars with non-evident caries lesions were selected. Two examiners independently evaluated 1,213 approximal surfaces in 126 children thorough visual inspection using the International Caries Detection and Assessment System (ICDAS), radiographic and LFpen methods. As reference standard, teeth were temporally separated with orthodontics rubbers for 7 days, and two examiners classified the surfaces in sound or with non-cavitated lesions. Concerning the occlusal surfaces, 407 primary molars in 68 children were assessed using the same methods. Results from the examinations with the ICDAS (consensus between two examiners) were considered as reference standard. Sensitivities, specificities, accuracies and utilities were calculated for each method alone and combined. The parameters were compared by McNemar test (p<0.05). Results: For proximal surfaces, ICDAS presented sensitivity=0.500; specificity=0.896; accuracy=0.656 and utility=86. Concerning the adjunct methods, the best strategy was ICDAS+radiographic (sensitivity=0.567; specificity=0.881; accuracy=0.691; utility=87); however, there were no significant differences compared to the results obtained with ICDAS alone. Regarding the occlusal surfaces, radiographic method presented poor sensitivity in detecting non-cavitated lesions (sensitivity=0.088; specificity=1.000; accuracy=0.164; utility=78). On the other hand, LFpen presented lower specificity and similar utility (sensitivity=0.687; specificity=0.813; accuracy=0.697; utility=79). These figures, however, do not probably overcome the performance of the ICDAS alone. Conclusion: Adjunct diagnostic methods do not offer benefits in detecting non-cavitated occlusal and proximal caries lesions in primary molars.