A relatively simple, quantitative approach is proposed to address a specific, important gap in the approach recommended by the USEPA Guidelines for Cancer Risk Assessment to address uncertainty in carcinogenic mode of action of certain chemicals when risk is extrapolated from bioassay data. These Guidelines recognize that some chemical carcinogens may have a site-specific mode of action (MOA) that is dual, involving mutation in addition to cell-killing induced hyperplasia. Although genotoxicity may contribute to increased risk at all doses, the Guidelines imply that for dual MOA (DMOA) carcinogens, judgment be used to compare and assess results obtained using separate "linear" (genotoxic) vs. "nonlinear" (nongenotoxic) approaches to low-level risk extrapolation. However, the Guidelines allow the latter approach to be used only when evidence is sufficient to parameterize a biologically based model that reliably extrapolates risk to low levels of concern. The Guidelines thus effectively prevent MOA uncertainty from being characterized and addressed when data are insufficient to parameterize such a model, but otherwise clearly support a DMOA. A bounding factor approach-similar to that used in reference dose procedures for classic toxicity endpoints-can address MOA uncertainty in a way that avoids explicit modeling of low-dose risk as a function of administered or internal dose. Even when a "nonlinear" toxicokinetic model cannot be fully validated, implications of DMOA uncertainty on low-dose risk may be bounded with reasonable confidence when target tumor types happen to be extremely rare. This concept was illustrated for a likely DMOA rodent carcinogen naphthalene, specifically to the issue of risk extrapolation from bioassay data on naphthalene-induced nasal tumors in rats. Bioassay data, supplemental toxicokinetic data, and related physiologically based pharmacokinetic and 2-stage stochastic carcinogenesis modeling results all clearly indicate that naphthalene is a DMOA carcinogen.Plausibility bounds on rat-tumor-type specific DMOA-related uncertainty were obtained using a 2-stage model adapted to reflect the empirical link between genotoxic and cytotoxic effects of the most potent identified genotoxic naphthalene metabolites, 1,2-and 1,4-naphthoquinone.Bound-specific "adjustment" factors were then used to reduce naphthalene risk estimated by linear extrapolation (under the default genotoxic MOA assumption), to account for the DMOA exhibited by this compound.3