Blood loss is a significant problem encountered in patients undergoing total joint arthroplasty, and is considered to be one of the factors affecting the outcome of the operation. Traditionally these patients have been treated with blood transfusions. The introduction of recombinant human erythropoietin (rHuEpo) into clinical practice enabled assessment of its effectiveness to decrease the allogeneic blood transfusion requirement (BTR), thus avoiding or minimizing transfusion-related complications. Fifteen patients undergoing total hip replacement (THR, 10 patients) and total knee replacement (TKR, 5 patients) in our institute (from January-April 1997), were studied. After signing an informed consent they received daily s.c. rHuEpo (100 IU/kg for those with hemoglobin (Hb) > 13 g/dl, 300 IU/kg for Hb < 13) during the 10 days prior to surgery and the 4 days following the operation. Allogeneic red blood cell (RBC) transfusions were given as needed. Hb levels were measured on days -10, 0, +1.3 and 7 of the procedure and the BTR was recorded. The results were compared with those of previous patients operated on from January-December 1996. Patients who were eligible for the study but refused to participate served as controls. The mean Hb level in the study group prior to rHuEpo administration (day -10) was 13.41 g/dl, similar to those of the control group (13.47 g/dl on day 0). However, the mean Hb levels in the rHuEpo treated patients on days 0, 1, 3 and 7 were 14.37, 11.09, 10.99, and 11.2 g/dl, respectively. This way compared with the levels of 13.47 (p = 0.016), 9.88 (p = 0.024), 9.60 (p = 0.004) and 9.97 g/dl (p = 0.007) in the control patients. The difference between the rHuEpo treated patients and the control patients was more significant among the THR patients than among the TKR patients. Of the 10 rHuEpo-treated THR patients, only a single patient required one allogeneic blood unit, as compared with 23 units transfused to the 30 control patients. None of the rHuEpo-treated TKR patients required blood transfusion as opposed to 4 units needed by the 11 control patients. In total, only one allogeneic blood unit was required by the study group which way calculated to an average consumption of 0.066 blood unit per person, compared with 27 blood units used by the 41 controls, i.e. 0.66 blood units per person (p < 0.001). In the patients treated, rHuEpo was very well tolerated with no adverse effects.