4,4′-MDA is classified as a genotoxic carcinogen based on numerous in vitro and animal data. The consequential assumption that a safe threshold does not exist is not only applied to 4,4′-MDA but also to its structural isomers and impurities 2,2′- and 2,4′-MDA in the absence of substance-specific data. This constitutes a problem in human risk assessments for all three substances as the inherent risks of 2,2′- and 2,4′-MDA and their contribution as impurities to that of 4,4′-MDA are essentially unknown. A comparative in vitro genotoxicity dataset consisting of the bacterial reverse mutation (Ames) test and the chromosomal aberration test in human lymphocytes (both performed according to the current OECD Guidelines) was generated for all three isomers. Furthermore, an in vitro gene mutation test in Chinese hamster ovary (CHO) cells (HPRT locus assay) was conducted with 2,4′-MDA. The results indicate differences regarding the genotoxic mechanism and potential, respectively, between the three structures and suggest that the no-threshold assumption for 4,4′-MDA may not be appropriate for 2,2′- and 2,4′-MDA.