The nature of the dose–response relationship for various in vivo endpoints of exposure and effect were investigated using the alkylating agents, methyl methanesulfonate (MMS) and methylnitrosourea (MNU). Six male F344 rats/group were dosed orally with 0, 0.5, 1, 5, 25 or 50mg/kg bw/day (mkd) of MMS, or 0, 0.01, 0.1, 1, 5, 10, 25 or 50 mkd of MNU, for 4 consecutive days and sacrificed 24h after the last dose. The dose–responses for multiple biomarkers of exposure and genotoxic effect were investigated. In MMS-treated rats, the hemoglobin adduct level, a systemic exposure biomarker, increased linearly with dose (r2 = 0.9990, P < 0.05), indicating the systemic availability of MMS; however, the N7MeG DNA adduct, a target exposure biomarker, exhibited a non-linear dose–response in blood and liver tissues. Blood reticulocyte micronuclei (MN), a genotoxic effect biomarker, exhibited a clear no-observed-genotoxic-effect-level (NOGEL) of 5 mkd as a point of departure (PoD) for MMS. Two separate dose–response models, the Lutz and Lutz model and the stepwise approach using PROC REG both supported a bilinear/threshold dose–response for MN induction. Liver gene expression, a mechanistic endpoint, also exhibited a bilinear dose–response. Similarly, in MNU-treated rats, hepatic DNA adducts, gene expression changes and MN all exhibited clear PoDs, with a NOGEL of 1 mkd for MN induction, although dose–response modeling of the MNU-induced MN data showed a better statistical fit for a linear dose–response. In summary, these results provide in vivo data that support the existence of clear non-linear dose–responses for a number of biologically significant events along the pathway for genotoxicity induced by DNA-reactive agents.