Recent evaluations of the regulatory in vitro genotoxicity tests have revealed that the mammalian cell tests are sensitive for the prediction of rodent carcinogenicity, but not very specific; they produce positive results for many non-carcinogens. Studies to investigate the potential causes of this poor specificity have focused on the high test concentrations of chemicals and the often associated high cytotoxicity. Others have sought to develop alternative in vitro assays that retain high sensitivity but also have higher specificity. One such test is the GADD45a-GFP "GreenScreen HC" genotoxicity assay, which uses elements of the cellular DNA damage response in a green fluorescent protein (GFP) reporter system to detect genotoxic hazard. Published validation studies for this assay have shown both high sensitivity and high specificity. Suggestions that the high specificity of the assay is attributable to the method of cytotoxicity estimation, relative cell density (RCD), and further that this method leads to overestimation of cytotoxicity and hence lower maximum test concentrations, are investigated in this study.In order to investigate whether high specificity of the GADD45a-GFP assay is a result of inaccurate estimation of cytotoxicity, RCD was compared with other cytotoxicity methods used in screening assays, as well as the cytotoxicity methods used in the regulatory in vitro mammalian genotoxicity tests. Results showed that RCD actually underestimates cytotoxicity in the GADD45-GFP assay and as a consequence, test data are collected up to and beyond the cytotoxicity-limited test concentrations reached in the regulatory tests. The use of RCD was found not to contribute to the high specificity of the GADD45a-GFP assay and indeed, its high specificity was retained regardless of the method used to estimate cytotoxicity.A largely overlooked element in the consideration of the poor specificity of in vitro mammalian genotoxicity assays is the supplementation of assays with exogenous metabolism systems, the most common of which is 'S9', a liver-derived preparation. S9 is used to enable the detection of compounds (pro-genotoxins) that undergo metabolic transformation in vivo resulting in the formation of genotoxic carcinogens. However, both the preparation and usage of S9 create a bias towards phase I metabolism, specifically the cytochrome P450 monooxygenases that can catalyse the formation of DNA-reactive species. The neglect of phase II metabolism (detoxification/conjugation) could potentially contribute to the poor specificity of the in vitro mammalian tests and this was explored in the study presented.A review of published data revealed that the prevalence of positive results amongst in vitro mammalian genotoxicity data generated in the presence of S9 was similar in collections of both rodent carcinogens and non-carcinogens. Thus, any increased sensitivity from using S9 is offset by decreased specificity for these tests. However, as inclusion of S9 affected sensitivity and specificity in equal measure it was concluded that S9 is not a confounding factor on the accuracy of the in vitro mammalian genotoxicity tests.