The development of time-dependent pharmacodynamic models in cancer chemotherapy has been extremely limited. A population approach was used to develop such a model to describe the effect of buthionine sulfoximine (BSO), via its active S-isomer (S-BSO), on glutathione (GSH) depletion in peripheral mononuclear cells. The Phase I trial utilized escalating doses of BSO, from 5 to 17 gm/m2, as a multiple infusion regimen. The population model consisted of a linear 2-compartment pharmacokinetic model coupled to an indirect response model. The indirect response model consisted of a GSH compartment with input and output rate processes that are modulated as a function of S-BSO and GSH concentrations. The model predicted the observed gradual depletion of GSH, a nadir at approximately 30 h after the last dose of BSO, and a return to baseline GSH levels. On the basis of an IC50 estimate of about 1.6 microM for inhibition of gamma-glutamylcysteine synthetase, the target enzyme of BSO, the population model predicted near identical GSH concentration time profiles over the dose range studied. Time-dependent pharmacodynamic models are seen as a powerful means to design dosing regimens and to provide a mathematical platform for mechanistic based models.