Despite the use of recombinant human erythropoietin (rh-EPO, epoetin) for more than a decade in treating renal anaemia, there is still considerable debate over optimal target haemoglobin (Hb) levels. Current European and North American guidelines that are based on decade-old trials aim for partial anaemia correction, with a subnormal target Hb concentration. More recent randomized clinical trials examining the effect of normalizing Hb levels have produced conflicting results. A study in the USA, in patients with existing congestive heart failure or ischaemic heart disease, showed an unexpected rise in cardiac mortality and haemodialysis access failure with higher Hb levels. In contrast, three other studies (in Australia, Spain and Canada) that normalized Hb levels in healthier dialysis patients observed improvements in quality of life and exercise capacity and a slower progression of left ventricular dilatation, without an unacceptable increase in the incidence of adverse effects. These studies indicate that, while higher Hb levels may be detrimental to patients with pre-existing cardiac disease, healthier patients benefit from normalized Hb levels. Thus, there is no clear scientific rationale for setting a single Hb target for all patients, and individualized treatment targets would appear to be a more logical and patient-centred approach.