The pharmacokinetics of recombinant human erythropoietin (Epo) were compared after mean single 99.1 U/kg intraperitoneal (IP), intravenous (IV), and subcutaneous (SC) doses in eight noninfected patients on peritoneal dialysis in a randomized, three-way, cross-over fashion. Continuous ambulatory peritoneal dialysis was performed in all patients on the days of the study. The IP dose was instilled into an empty peritoneum; total dwell time was 10 hours (4 hours dry, 6 hours with 2 L of peritoneal dialysis fluid). Blood samples were collected for 96 hours following IP and SC Epo, and for 72 hours following IV Epo. For the IP dose, a 10-hour effluent dialysate sample was collected to determine Epo recovery. Enzyme immunoassay was used for Epo analysis. The mean apparent volume of distribution was 0.05 L/kg, equivalent to 4.5% of total body weight; the mean total body clearance was 0.08 mL/min/kg. All eight patients exhibited multiexponential decay in serum Epo concentrations following IV Epo. Absorption of IP Epo was significantly greater than previous reports, presumably due to its administration into a dry peritoneum. The maximum concentrations following the IP and SC doses were nearly identical, but amounted to only 5% of the maximum concentrations for the IV dose. Subcutaneous Epo took nearly twice as long as IP Epo to achieve peak serum concentrations (17.1 ± 5.0 hours v 9.4 ± 1.9 hours). Compared with the IP route, the SC dose achieved a higher area under the serum concentration time curve from time 0 to 96 hours (AUC0.96; P = 0.0001). In addition, SC Epo was significantly more bioavailable than IP Epo (21.8% ± 15% v 11.4% ± 6.6%; P = 0.0001). Furthermore, 25.1% ± 11% of the IP dose was recovered in the 10-hour dialysate effluent. During the 96 hours of the study, 77.2% ± 14% of the SC dose was not absorbed. Compared with the IP dose, there was a larger percentage of unaccounted SC Epo dose (63.5% ± 13% v 77.2% ± 14%; P <0.05). In conclusion, despite its modest bioavailability, the pharmacokinetics of Epo following administration into a dry peritoneum suggest that the IP route may have potential clinical usefulness.