In order to optimize the treatment of anemia in uremic patients the pharmacokinetic and pharmacodynamic properties of recombinant erythropoietin (r-Epo) were studied after i.v., s.c. and i.p. administration. Both in healthy volunteers and in patient with chronic renal failure, the half-life of r-Epo after i.v. administration was short (about 6 h) in comparison with the commonly used dosing interval of 3 to 7 days. Hence, the minimum serum concentration during a dosing interval is expected to be less than 1% of the peak concentration. Absorption following a s.c. dose was slow, resulting in a markedly different concentration-time profile in comparison to i.v. dosing. The half-life was about 25 h and only approximately 25% of the given dose reached the systemic circulation. As a result of differences in concentration-time profiles, higher trough concentrations during s.c. dosing intervals may be expected in comparison to those occurring after i.v. dosing. When r-Epo was given i.p. (diluted in dialysate) the extent of systemic absorption depended on the dwell time in the peritoneal cavity. A long administration time was required to absorb an amount of r-Epo predicted from s.c. studies to be adequate to achieve the desired clinical effect. In spite of reduced bioavailability, s.c. treatment did not require higher r-Epo doses than i.v. treatment to maintain the desired hemoglobin concentration. On the contrary, a trend to a requirement for lower doses was detected. The pharmacodynamic and pharmacokinetic results strongly indicate a more efficacious concentration-time profile following s.c. administration. Since s.c. dosing also allows self-administration the use of this administration route is recommended. To simplify the treatment of anemic patients with r-Epo, a model was developed to predict the required weekly s.c. dose. To facilitate the use of this model a nomogram was constructed.