In the assessment of risk, several factors affect predictions: selection of reactive agent, selection of tumor incidence data, modeling of dose, scaling across species, adjustment for differences in duration and frequency of exposure, and selection of the most suitable risk extrapolation model. If the end points, exposure regimen, and the model for risk extrapolation are constant, then the review of dose paradigms will illustrate the effect of dose modeling on risk, since by definition the reactive agent is the primary carcinogen. The response incidence in lifetime inhalation bioassays of two primary carcinogens, ethylene oxide and formaldehyde, was used with different dose paradigms to estimate risk from maximum lifetime occupational exposures. The dose paradigms that will be considered include: concentration, concentration time product, retained dose, integrated blood concentration, and tissue exposure. (Copyright (c) 1989 Health Physics Society.)