The relative mutagenicities of substituted N-nitroso-N-benzylmethylamines have been reexamined from a quantitative structure-activity relationship point of view. Most of the compounds were mutagenic toward Salmonella typhimurium TA 1535 with Aroclor-induced male hamster liver S9 activation. The dose-response data were subjected to a multiple linear regression equation calculated in a stepwise manner, which found that the differences in mutagenicities could be explained primarily by differences in the three-bond path molecular connectivity index, with smaller contributions from sigma and pi. Moreover, a polynomial regression analysis showed that the maximum mutagenicity could be explained by an optimal amount of electron withdrawal by the substituent which would cause a weakening, or activation, of the methylene C-H bond. The possible relevance of these observations to carcinogenesis is discussed.