The US Environmental Protection Agency (USEPA) is currently conducting a toxicological review of vanadium pentoxide (V2O5). As part of that effort, the Agency will need to address the fact that while a National Toxicology Program (NTP) chronic inhalation bioassay of V2O5 produced clear evidence of treatment-related lung tumors in both male and female B6C3F1 mice, neither of these responses were dose-related across the groups exposed to 1, 2, and 4mg/m(3). While lung tumor incidence was significantly elevated in all three exposed groups relative to that in the control groups, it was essentially flat across them. Herein we report results from computing poly-3-adjusted Cochran-Armitage trend test statistics with and without inclusion of the lung tumor incidence data from control group mice. These results confirm the absence of any significant dose-related effect on mouse lung tumor incidence in the study groups exposed to V2O5. We also considered two estimates of area under the vanadium lung burden versus time curve as plausible alternative dose metrics to the V2O5 chamber concentration. However, these alternative dose metrics were so highly correlated with the V2O5 chamber concentration (r=0.998) that nothing is to be gained from their use in place of the V2O5 chamber concentration in attempts to perform dose-response modeling of the tumor incidence or unit cancer risk computations. At the present time, there is no scientific basis to support linear (or nonlinear) extrapolations of estimated cancer risks to V2O5 exposure levels below 1mg/m(3). Additional tumor data at multiple V2O5 concentrations lower than 1mg/m(3) are required to support such extrapolations.