We have retrospectively analysed sera from 52 already sensitized uraemic patients collected over 1 year and compared erythropoietin (EPO)-treated with non-EPO-treated patients. Significantly fewer (P<0.01) patients (33%) on dialysis because of the rejection of their kidney grafts received EPO than patients on dialysis because of underlying kidney disease (71%). EPO treatment reduced the number of additional blood transfusions, since 3/28 EPO-treated but 12/24 non-EPO-treated patients were given blood (P<0.05). Among the EPO-treated patients, 64% showed a loss of panel-reactive antibodies (PRA), as measured by the micro-lymphocytotoxic technique, while only 12.5% of the non-treated patients showed a loss of PRA (P<0.01). In the subgroup of transplanted patients, PRA loss was only found among the EPO-treated patients, but their number was small (P<0.05). The class, subclass and specificities of the antibodies, as determined by FACS (flow cytometry) analyses, showed no distinct differences between EPO- and non-EPO-treated patients. The differences were significant between transfused and previously transplanted patients.