Sotorasib (Lumakras™) is a first-in-class, non-genotoxic, small molecule inhibitor of KRAS G12C developed as an anticancer therapeutic for treatment of patients that have a high unmet medical need. Anticancer therapeutics are considered out of scope of ICH M7 guidance for control of mutagenic impurities; however, based on ICH S9 Q&A, mutagenicity assessments are needed for impurities that exceed the qualification threshold, consistent with ICH Q3A/B, and non-mutagenic drugs. Here, we carried out hybrid-based mutagenicity assessment of sotorasib drug substance (DS) impurities using in silico quantitative structure-activity relationship (QSAR) modelling and Ames tests (for in silico positive mutagens). We encountered contradictive mutagenicity results for 2 impurities (Beta-Chloride and PAC). PAC was negative initially by QSAR but positive in a GLP full plate Ames test and Beta-Chloride was positive by QSAR, negative in a non-GLP micro-Ames but positive in a GLP full plate Ames assay. Root cause analyses identified and characterized mutagenic contaminants, 3-chloropropionic acid in batches of Beta-Chloride and 3-chloropropionic acid and Chloro-PAC in batches of PAC, used in initial GLP full-plate Ames tests. Significant reduction of these contaminants in re-purified batches resulted in no induction of mutagenicity in follow-up GLP micro-Ames tests. In summary, root-cause analyses led to accurate mutagenicity assessment for sotorasib DS-associated impurities.