“…Prediction of mutagenicity of reagent-sized ArNH 2 fragments is vital for the success of drug discovery programs, and special attention has been devoted to this issue in the literature. Currently, the mutagenicity of ArNH 2 can be predicted by three types of computational methods, which are based on either statistics, structural alerts, or mechanistic considerations. − ,− Despite the abundance of experimental data and the availability of computational approaches and relevant databases, the prediction of mutagenicity of ArNH 2 still represents a significant challenge to computational and theoretical chemistry. ,,,,, Quantitative structure–mutagenicity relationships usually suggest a variety of factors that increase mutagenic potency of ArNH 2 : increase of lipophilicity, high energy of the highest occupied molecular orbital (HOMO), low energy of the lowest unoccupied molecular orbital (LUMO), low “chemical hardness” (LUMO–HOMO), high “chemical softness”, stability of ArNH + (reactive electrophilic metabolites of ArNH 2 ), negative charge density at the exocyclic nitrogen of ArNH + , presence of a pyridine-like nitrogen (i.e., aromatic nitrogen with sp 2 lone pair) in the α-position to the NH 2 group, and size of the aromatic system. ,,,− ,,,,− Although these descriptors reflect an array of steps in the mutagenic activation of ArNH 2 and are important for mutagenic activity, ,,,, their ability to discriminate between mutagenic and nonmutagenic compounds , or to describe mutagenic potency in diverse sets of active mutagens ,,,,,, is limited. Thus, high lipophilicity, high energy of HOMO, and even stability of ArNH + turn out to be unimportant to define high mutagenic potency or to discriminate a propensit...…”