The study was performed to measure the impact of testosterone (T) administration on circulating levels of 5alpha-dihydrotestosterone (DHT). Group 1 (32 men; mean age 61 years; mean T 6.9 +/- 1.9 nmol l(-1)) were treated for 15 months with long-acting T undecanoate. Group 2 (23 men, mean age 60 years, mean T 7.6 +/- 2.0 nmol l(-1)) were treated for 9 months with T gel. Plasma T and DHT were measured before and after 9 months T administration. In the men treated with T undecanoate plasma T and DHT were also measured after 12 and 15 months. Before T administration, plasma DHT ranged from 0.39 to 1.76 nmol l(-1) (0.30-1.90 nmol l(-1)). Mean DHT declined upon T administration from 0.95 +/- 0.50 to 0.55 +/- 0.30 nmol l(-1) (P < 0.05). With an arbitrary cut-off at 0.60 nmol l(-1), all 21 values of DHT > 0.60 nmol l(-1) had fallen from 1.29 +/- 0.50 to 0.70 +/- 0.60 nmol l(-1) (P < 0.01). Below this cut-off point 13 values rose and 21 fell upon T administration. Below this cut-off point values on average declined from 0.39 +/- 0.12 to 0.30 +/- 0.14 nmol l(-1) (P < 0.05). The study revealed that in a cohort of elderly men with subnormal plasma T levels plasma DHT levels declined upon T administration when they were in the higher range of normal (>0.6 nmol l(-1)), with a profound shift of DHT/T ratios presumed to be an indicator of a reduced 5alpha-reductase activity. Below plasma DHT levels of 0.6 nmol l(-1), responses of plasma DHT to T administration varied.