It is not known whether recombinant human erythropoietin has a direct, clinically apparent pressor effect in hemodialysis patients or whether hypertension developing or aggravated in these patients merely reflects increased hematocrit. We compared blood pressure after three different methods of partial correction of anemia in hemodialysis patients with similar baseline hematocrits (erythropoietin n = 12, intravenous iron alone n = 10, androgens n = 9). Shortly after the start of treatment and with a minimally increased hematocrit, the need for antihypertensive medication increased in the erythropoietin group. No such pressor effect was observed with iron or androgens. These data suggest a direct hypertensive effect of erythropoietin in some patients on hemodialysis.