Background In vivo reflectance confocal microscopy significantly improves melanoma diagnosis as compared to clinical/dermoscopic examination alone. Several confocal criteria have been described allowing to differentiate melanoma from nevi; by combining different criteria, three pure confocal scores (Pellacani 2005, Segura 2009 and Pellacani 2012) and one mixed dermoscopic/confocal score (Borsari 2018) were constructed. Objective Our aim was to externally validate and compare the performance of these confocal scores. Methods We retrospectively enrolled excised melanocytic lesions which underwent confocal examination in a 2‐year period. Lesions located on the face and acral sites were excluded. Both dermoscopic and confocal criteria considered in the four scores were evaluated by experts. Subsequently, specificity and sensitivity levels for each score were calculated, together with the positive and negative predictive values and likelihood ratios; also, receiver operating characteristic curves were constructed. Results A total of 389 patients with 422 lesions were retrospectively enrolled, of which 162 (38.4%) were melanomas and 260 (61.6%) were nevi (189 common and 71 Spitz/Reed nevi). The highest sensitivity levels were recorded for Segura 2009 with cut‐off ≥−1 (92.0%), while Pellacani 2005 with cut‐off ≥5 achieved the highest specificity (69.6%). The score by Borsari et al. showed the highest levels of positive and negative predictive values (59.8% and 91.5%) and likelihood ratios (2.4 and 0.1) as well as the highest area under the curve values (0.76; 95% CI 0.72–0.81; P < 0.001). Conclusions High levels of accuracy were found for each of the four considered scores. No differences were found among scores in confirming melanoma diagnosis when positive; however, the score by Borsari 2018 was the best in excluding melanoma diagnosis when negative.