Capecitabine has been developed as an orally administered tumor selective fluoropyrimidine for use in the treatment of breast and colorectal cancer. The metabolic pathway for capecitabine includes 5'-deoxy-5-fluorocytidine (5'-DFCR) and 5'-deoxy-5-fluorouridine (5'-DFUR), which is then converted to the pharmacologically active agent 5-fluorouracil (5-FU). A previous analysis showed that systemic exposure to 5'-DFUR and alpha-fluoro-beta-alanine (FBAL), a catabolite of 5-FU, was predictive of dose limiting toxicities. Therefore, a multi-response population pharmacokinetic (PK) model for the description of plasma concentrations of 5'-DFUR, 5-FU and FBAL following oral administration of capecitabine was developed using NONMEM. PK data from a bioequivalence study in 24 patients with various solid tumors were used to develop the PK structural part of the population PK model. The 5'-DFUR, 5-FU and FBAL plasma concentrations were described by a linear disposition PK model with first order absorption and lag time. Sparse plasma concentration data from 54 phase II breast cancer patients were added to the bioequivalence data and the influence of covariates on the apparent oral clearances of 5'-DFUR, 5-FU and FBAL and on the apparent volume of distribution of FBAL was investigated. This was conducted by including all significant (p < 0.05) single covariate-PK parameter pairs in the full PK model, followed by one by one deletion (p < 0.001) from the population model. Statistically significant effects were found for the influence of gender, body surface area and total bilirubin on 5'-DFUR clearance and the influence of creatinine clearance on FBAL clearance. However, none of these effects were considered to have clinical relevance.