Acute normovolaemic haemodilution (ANH) is used to avoid perioperative blood loss and consists of the withdrawal of whole blood just before or just after anaesthesia induction and its simultaneous replacement by synthetic colloids and crystalloid solutions. In an attempt to improve the efficiency of this technique while at the same time avoiding cardiovascular complications, we set up a pilot study to test the association of rHuEpo/ANH during elective surgery for total hip replacement. Five patients (3 males, 2 females) were included in this study. The amount of whole blood drawn was 3 x 450 ml from the men and 2 x 450 ml from the women. Before blood was taken, the mean increase in haemoglobin was 1.2 +/- 0.9 g/dl and mean increase in reticulocytes 106 +/- 34 G/l. No patient received homologous transfusion during the perioperative period; 3 patients received the totality of predonated blood and one patient 2 of the 3 units taken. The mean fall in haemoglobin at day 1 post-surgery was 3.6 g/dl. In conclusion, the stimulation of erythropoiesis by rHuEpo in the pre-surgery phase led on average to a 1 g/dl gain in haemoglobin, permitting an isovolaemic withdrawal of 900 to 1350 ml of blood depending on body weight without the development of severe anaemia. It was thus possible to perform total hip replacement in all the patients without homologous blood support and with a post-surgery haemoglobin value of > 10 g/dl. This protocol should be further tested in a prospective randomised study (rHuEpo versus placebo) in order to assess the real benefit of rHuEpo.