The prospect of a shortage of blood for transfusions, increasing awareness of the adverse effects of transfusions, and the availability of human recombinant erythropoietin (rHuEPO) have stimulated interest in the pathogenesis of the anaemia of intensive care unit (ICU) patients. As in the anaemia of chronic illness or chronic renal failure (CRF), the anaemia of ICU patients is a multifactorial process. Blood loss, inappropriately low erythropoietin production, reduced red cell lifespan, reduced iron availability, and inhibition of erythropoiesis by cytokines all contribute to the anaemia of critical illness, although the contributions of the various elements differ depending on the disease aetiology. Evidence is accumulating that use of rHuEPO can induce stimulation of erythropoiesis in critical illness, but at doses that are usually several-fold higher than those used to define resistance to rHuEPO in the current guidelines for the management of anaemia in CRF. Available data suggest that these high doses are well tolerated, at least in the short term. These observations, as well as demonstrating the potential benefits of rHuEPO therapy in critically ill patients, have practical implications for non-ICU patients with CRF who do not respond sufficiently to the usual doses of rHuEPO. Although the risk-benefit ratio relationship for very high doses of rHuEPO needs further consideration, demonstration of rHuEPO efficacy in critical illness should result in a re-evaluation of the 'dose-response relationship' for rHuEPO in patients with less acute and severe illness, including CRF patients hyporesponsive to current dosing regimens.