Immunoreactive serum erythropoietin (EPO) was measured in anemic and non-anemic patients with acquired non-severe aplastic anemia (AA; n = 22) and myelodysplastic syndromes (MDS; n = 31) receiving or not androgens to examine the effect of androgen therapy and anemia on EPO levels in these disorders. Soluble transferrin receptor (TfR) and absolute reticulocyte count (ARC) were also assayed in order to evaluate erythropoietic activity. AA and MDS patients were stratified for anemia and androgen treatment as follows: 12 untreated anemic patients; 17 anemic patients during androgen therapy; 14 non-anemic patients without any treatment (>1 year); and 10 non-anemic patients on androgen therapy. Although EPO levels in non-anemic patients were significantly higher than in healthy controls (n = 29) no statistically significant differences in Hb and EPO values were found between non-anemic patients receiving or not androgen therapy. In the linear regression analysis between Hb and log EPO concentration, no statistically significant differences in the slopes between untreated and androgen-treated anemic groups nor between both groups and patients with iron deficiency anemia (n = 23) were observed. However, the y intercept (log EPO) of regression line was significantly higher in androgen-treated anemic patients than in the androgen therapy-free anemic group. Serum TfR levels were higher in treated than in untreated anemic patients, whereas ARC was not different between both groups. These data seemingly indicate that (1) androgens at pharmacological doses do not increase serum EPO levels in non-anemic AA and MDS patients, and (2) in patients with AA and MDS, androgen-driven EPO stimulation is appreciably enhanced by anemia. Am.