This multicentre study in 142 transfusion-dependent patients with chronic renal failure maintained by haemodialysis was performed to establish the appropriate dose regimen of rHuEpo and define its long-term safety profile. Only one of 132 patients eligible for efficacy analysis did not achieve the haemoglobin target of greater than or equal to 10 g/dl; this particular patient had folate deficiency and overt hyperparathyroidism. Regular blood transfusions were no longer necessary in any patients, however five patients needed blood transfusions only once, not due to rHuEpo failure: two for iron deficiency and three for intercurrent disease. In parallel with the haemoglobin increase a statistically significant improvement in quality of life scores was observed. The weekly dose required to maintain median haemoglobin between 10 and 10.5 g/dl for 1 year (n = 79) was 200-225 U/kg, applied as two or three i.v. injections. Mean serum ferritin decreased from 1900 to 1300 ng/ml and transferrin saturation from 60% to 30%; this feature was associated with statistically significant decrease of pre-study elevated liver enzymes. The treatment had no untoward effect on the outcome of renal transplantation (n = 24). Of the 56 patients who experienced hypertensive episodes during rHuEpo therapy, 47 had a history of hypertension and nine had not. The patient incidence during the first 3 months was 28.9% and fell markedly to 4% after 1 year. Only two hypertensive episodes could not be controlled and the patients dropped out. Seizures occurred in 11 patients, most of them during early treatment; annualised incidence during the first 3 months was 7.78 per year vs 2.07 per year for seizures beyond 3 months treatment. Clinical presentation, patients' history, haemoglobin pattern, BP recordings, brain scan, and EEG indicated that the pathophysiology is multifactorial, with emphasis on rate of haemoglobin increase. Therefore a smooth haemoglobin increase rate, induced by a conservative starting dose regimen (50 U/kg thrice weekly) is recommended, to allow the circulation to adapt to changes in haematocrit/viscosity and O2 delivery. The majority of the observed adverse reactions were related to rHuEpo's therapeutic effect, i.e. increase the haematocrit. The side-effects are therefore largely predictable and can be successfully managed.