Our earlier analyses have suggested an apparent threshold dose-response for ethylnitrosourea-induced specific-locus mutations in treated spermatogonia of the mouse to be due to a saturable repair process. In the current study a series of fractionated-treatment experiments was carried out in which male (102 x C3H)F1 mice were exposed to 4 x 10, 2 x 40. 4 x 20 or 4 x 40 mg ethylnitrosourea per kg body weight with 24 h between applications; 4 x 40 mg ethylnitrosourea per kg body weight with 72 h between dose applications; and 2 x 40, 4 x 20 and 4 x 40 mg ethylnitrosourea per kg body weight with 168 h between dose applications. For all experiments with 24-h intervals between dose applications, there was no effect due to dose fractionation on the observed mutation rates, indicating the time interval between dose applications to be shorter than the recovery time of the repair processes acting on ethylnitrosourea-induced DNA adducts. In contrast, a fractionation interval of 168 h was associated with a significant reduction in the observed mutation rate due to recovery of the repair process. However, although reduced, the observed mutation rates for fractionation intervals of 168 h were higher than the spontaneous specific-locus mutation rate. These observations contradict the expectation for a true threshold dose response. We interpret this discrepancy to be due to the differences in the predictions of a mathematical abstraction of experimental data and the complexities of the biological system being studied. Biologically plausible explanations of the discrepancy are presented.