There is substantial mutagenicity data on acyl/sulfonyl halides available in the public domain and these data are the basis for many in silico models of mutagenicity (e.g. DEREK Nexus and Leadscope). A review of these data indicates that the perceived mutagenic potential of this class of compounds is based on a number of non-reproducible positive findings in the bacterial mutagenicity assay and positive bacterial mutagenicity data on a series of compounds where formation of reactive halodimethylsulfides (HDMS) in DMSO may have compromised the interpretation of the Ames data (HDMS are typically mutagenic). The only genuine mutagenic, genotoxic and carcinogenic compound within the 50+ acyl/sulfonyl halides described herein, was dimethylcarbamic chloride, which is appropriately considered to be a potential human carcinogen. Some in silico systems, such as Derek Nexus, contain rules which detail that the activity of this class should be considered a false positive flag for mutagenicity and ideally any in silico structure activity rules for mutagenicity in other systems should be likewise addressed. The data presented here supports the view that these alerts should currently be interpreted as a false positive flag for mutagenicity and the entire class viewed as low concern from a mutagenicity perspective. The formation of these reactive halodimethylsulfides is an example of the classical Pummerer rearrangement.The chemical reactivity of this class of compounds also supports the contention that they are of limited concern from a mutagenic and carcinogenic impurity risk perspective when used in the synthesis of drug products. They can be expected to rapidly purge from any reaction sequence with generic predicted purge factors in the range 1x10 3 -3x10 5 /stage and hence should be effectively eliminated at the stage of introduction.We would therefore recommend avoiding DMSO as solvent for mutagenicity tests with acyl/sulfonyl halides due to the potential for false positive results arising from DMSO reaction products, which are not relevant in aqueous, physiological conditions. Furthermore, as indicated by the Ames test data for mesyl chloride/2-fluorobenzoyl chloride, even non-DMSO organic solvents may not be appropriate for certain members of this class (acyl/sulfonyl halides), suggesting that they may not be amenable to adequate testing in the Ames assay.