To compare the erythropoietic effects of nandrolone decanoate, testosterone enanthate, oxymetholone, and fluoxymesterone, we performed a randomized clinical trial in patients with anemia who were receiving maintenance hemodialysis (the women were not given testosterone enanthate). After a control period of at least two months, patients received one of the drugs for six months and then returned to control status; a second and third drug were administered in a similar fashion. Seventy-seven patients completed the first drug period, 56 the second, and 35 the third. The response to nandrolone and testosterone enanthate, the two drugs given by injection, was clearly superior to the response to oxymetholone or fluoxymesterone, given by mouth, in terms of the percentage of patients responding and the mean rise in hematocrit. Approximately half the patients had an increase of at least 5 percentage points in hematocrit after an injectable androgen was given; more than half the women responded. Patients who required transfusions regularly and those who had bilateral nephrectomies did not respond.