This study describes a pharmacokinetic (PK) model to account for serum recombinant human erythropoietin (rHuEpo) concentrations in healthy volunteers following intravenous (IV) and subcutaneous (SC) dosing; it also characterizes the pharmacodynamics (PD) of SC rHuEpo effects on reticulocytes, red blood cells (RBC), and hemoglobin (Hb) in blood. Data were obtained from 4 clinical studies carried out in healthy volunteers. Epoetin alfa (rHuEpo) was administered as 5 single IV doses ranging from 10 to 500 IU/kg, as 8 single SC doses ranging from 300 to 2400 IU/kg, and as 2 multiple SC dosage regimens (150 IU/kg/3 times a week [tiw] and 600 IU/kg/wk). A dual-absorption rate model (fast zero-order and slow first-order inputs) with nonlinear disposition characterized the PK of SC rHuEpo. A high K(m) value was obtained indicating that clearance was mildly nonlinear. Absorption was slow (t(max) approximately 24 hours), and the bioavailability of SC rHuEpo increased with dose (ranging from 46%-100%). A catenary cell production and loss model with a feedback down regulation component was used to fit the reticulocyte data yielding estimates of the stimulatory capacity (S(max)), sensitivity (SC(50)), and life span parameters. These parameters were used for simulations of RBC and Hb profiles. An SC(50) of 27 to 61 IU/L was estimated indicating that low physiological plasma rHuEpo concentrations were sufficient to produce pharmacological effects. No marked sex-dependent differences in clinical responses to rHuEpo therapy were found despite baseline differences. Realistic pharmacokinetic and physiological models accounted for clinical responses from a wide array of dosing conditions with rHuEpo. The rationale for greater efficacy of SC administration of rHuEpo compared to IV was ascertained.