Objective: To perform a systematic review and meta-analysis of the performance of different methods for detecting carious lesions in permanent and primary teeth, considering all types of tooth surface. Methods: Two reviewers searched PubMed, Embase, Scopus and other sources up to November 2020 to identify published and nonpublished studies in English. We focused on three caries detection methods: visual inspection (VI), radiographic (RX) and fluorescence-based (LF). We included studies investigating at least one of these methods which (a) assessed the accuracy of the method in detecting caries lesions; (b) considered occlusal, proximal or free smooth surfaces in primary or permanent teeth; (c) used a reference standard other than one of the three methods; and (d) reported data on sample size and accuracy. Multilevel analyses, meta-regressions and comparisons of bivariate summary receiver operating characteristics curves were undertaken. Results: Two hundred and forty manuscripts from 14 129 articles initially identified met the inclusion criteria. VI was better than RX on occlusal surfaces at all caries lesion thresholds and proximal surfaces of permanent teeth only at all lesion thresholds in laboratory setting. LF was slightly better than VI for advanced lesions on occlusal surfaces of permanent teeth in the clinical setting and for all lesions on proximal surfaces of permanent teeth in the laboratory setting. Still, LF was worse than VI for advanced occlusal lesions in permanent teeth in the laboratory setting. Although LF showed slightly better performance than VI with advanced lesions, the latter had significantly higher specificity than other methods in all settings. Conclusion: Visual caries detection alone is adequate for most patients in daily clinical practice regardless of tooth type or surface.