The use of social tasks in the neuropsychological assessment of the behavioral variant of frontotemporal dementia (bvFTD) is at present not required by diagnostic guidelines, despite extensive literature shows relevant social cognitive dysfunctions in such patients. In this systematic review, we explored the clinical maturity of social cognition measures in the diagnosis of bvFTD. Papers were selected according to the PRISMA guidelines by searching the PubMed and Medline databases. Only papers reporting indices of diagnostic accuracy and/or sensitivity/specificity in classifying bvFTD from controls or from other relevant diseases were considered. Quality of evidence was assessed through QUADAS-2. Among the 663 articles entered in the paper selection only 14 papers were eligible for the scope of the present review and showed an overall moderate-to-low quality. The major risk of bias was the lack of pathological confirmation. The evaluation of the accuracy of social cognition tasks in bvFTD detection compared to normal controls, as well as in the discrimination with Alzheimer's disease and psychiatric patients, is mainly focused on emotion recognition and theory of mind. However, the use of different cognitive measures, variable task formats and the limited normative data hamper study comparability. Although literature seems to suggest that emotion recognition and ToM tasks could be the best choice to ensure a high diagnostic accuracy in clinical settings, further comparative studies are required and no recommendation concerning the use of a specific social task in bvFTD diagnosis can be currently provided.