In order to investigate differential effects of androgens on erythropoiesis, 55 men with clincally and biochemical confirmed hypogonadism were randomly assigned to 4 groups receiving different forms of androgen substitution: Mesterolone (MES) 100 mg/d, testosterone undecanoate (TU) 160 mg/d, testosterone enanthate (TE) 250 mg i.m./21 days or 1200 mg crystalline testosterone (TPEL) subcutaneously implanted at study begin. Previous testosterone medication had been suspended at least 3 months prior to study begin. Testosterone (T), dihydrotestosterone (DHT), hemoglobin (HB) and hematocrit (HC) were assessed before, during and after substitution of androgens. MES did not increase serum T and TU raised average T levels during substitution to 5.7 +/- 0.3 nmol/l, thereby doubling baseline concentrations. TE resulted in a 6fold increase of baseline T yielding 13.5 +/- 0.7 nmol/l and TPEL increased serum T 8.5fold to 23.2 +/- 1.1 nmol/l. Average DHT levels during substitution were 4.3 +/- 0.2 (MES), 3.3 +/- 0.2 (TU), 4.0 +/- 0.4 (TE) and 5.5 +/- 0.4 (TPEL) nmol/l. The groups receiving TPEL, TU or TE showed a significant rise of HB and HC compared to baseline, whereas in the MES group these parameters did not change significantly. MES increased HB by 5.6 +/- 1.8 g/l, TU by 12.7 +/- 2.8 g/l, TE by 21.1 +/- 2.6 g/l and TPEL by 21.7 +/- 4.0 g/l. HC was raised by 1.8 +/- 0. 4% in the MES group, 3.9 +/- 1.1% in the TU group and 6.4 +/- 0.9% and 6.5 +/- 1.6% in the TE and TPEL groups, respectively. Except for 1 subject in the TPEL group, the HB and HC stayed within the normal limits. We conclude that, T, but not DHT, stimulates erythropoiesis in a dose dependent manner. T levels within the low normal range for men are required for maximal stimulation of erythropoiesis.