Starting in 1998, the number of pure red-cell aplasia (PRCA) cases in patients treated with recombinant human erythropoietin (rHuEPO) increased dramatically. Most cases were observed in patients treated with epoetin alfa produced outside the United States. The peak was observed in 2002; since then, the PRCA incidence has declined. Many factors are likely to have contributed to this up-surge. The molecular structure of the various epoetins and patient characteristics do not seem to play a major role. The route of administration holds some importance, because most PRCA patients received rHuEPO subcutaneously. The peak of PRCA cases overlapped with the removal of human serum albumin from the Eprex formulation (Janssen-Pharmaceutica NV, Beerse, Belgium), for which polysorbate 80 and glycine were substituted. Polysorbate 80 may have increased the immunogenicity of Eprex by eliciting the formation of epoetin-containing micelles or by interacting with leachates released by the uncoated rubber stoppers of prefilled syringes. Compared with the previous formulation, the polysorbate 80 formulation has lower stability, making it more susceptible to stress conditions such as insufficient attention to the cold chain. This situation could facilitate protein denaturation or aggregate formation. Uncoated rubber stoppers were replaced with coated stoppers, and the cold chain was reinforced; the Eprex formulation has remained unchanged. Even though the incidence of PRCA returned to very low levels, discriminating the cause-effect relationship of a single action is difficult, given that all occurred with a similar chronology, and that PRCA develops after a relatively long exposure period. Careful observation of future trends of new PRCA cases is thus mandatory.