Biologically based dose-response (BBDR) models comprise one way to incorporate mechanistic information into a dose-response assessment to be used for risk assessments. The chemotherapeutic drug 5-fluorouracil (5-FU) has been used as a prototypic compound for the construction of a BBDR model for developmental toxicity. Previous work has provided data and a general mechanistic framework for the developmental toxicity of 5-FU when it was administered to pregnant rats subcutaneously on gestation day 14. In this paper, a mathematical model relating maternally administered treatment with 5-FU to embryonal thymidylate synthetase inhibition and thymidylate synthetase inhibition to various measures of deoxyribonucleotide triphosphate (dNTP) pool perturbation is developed, and parameters are estimated using the data collected. The strategy used was to develop semi-empirical submodels for each of the intervening steps, and to estimate model parameters from previously described data. The models developed predict that there is no practical threshold for dNTP pool perturbation; that is, even minimal doses of 5-FU should result in some perturbation of dNTP pools. In particular, the relationship between dNTP pool perturbation and fetal weight deficit suggests that if there is a biological threshold for the effect of 5-FU on fetal weight, the responsible repair or compensation mechanism must be downstream of dNTP pool perturbation, and saturable at 5-FU doses lower than 10 mg/kg (the lowest dose examined for developmental effects in these studies).