The oncogenic potential of isoprene as affected by concentration, length of daily exposure, and weeks of exposure over the life-span of the animal, as independent variables, was evaluated. Ten groups were exposed for 8 h/day, 5 days/week as follows (ppm-weeks): 0-80, 10-80, 70-40, 70-80, 140-40, 280-20, 280-80, 700-80, 2200-40, 2200-80. Two groups were exposed for 4 h/day: 2200-20, 2200-80. Groups were held until 96 or 105 weeks on study. The concentration x time (duration of exposure) values provided a series of theoretically equivalent exposure hazards. There was an exposure-related increased incidence of liver, lung, Harderian gland and forestomach tumors, and hemangiosarcomas and histiocytic sarcomas. The LOEL appeared to be 70 ppm. These results are similar to the profile of tumors seen in 1,3-butadiene (BD)-exposed mice without the early onset of T-cell lymphoma as seen with BD. Isoprene appears to be about one order of magnitude less potent than BD in mice. Statistical analyses indicated that the product of isoprene concentration, and length/duration of exposure was not a sufficient basis for predicting tumor risk at any site. Extrapolation of tumor probability between the high and low doses based on cumulative exposure was not appropriate and could not be justified by statistical models. A threshold effect level and strong nonlinearities with respect to concentration appeared to exist for tumor development in this study.