This study offers thorough scrutiny of the methodological problems that affect the studies performed in this new and underexplored field of computer-assisted melanoma diagnosis. The authors discuss the limitations of the studies, including selection and verification bias, unclear inclusion and exclusion criteria, the lack of use of an independent test set, and the absence of interinstrument and intrainstrument reproducibility assessment. The authors conclude that the results of all these studies need to be interpreted with caution, and they suggest a list of minimum requirements for future studies to ensure better methodological quality.