To test the reliability of a single plasma testosterone (T) or dihydrotestosterone (DHT) level, respectively, as a parameter of the long term hormonal milieu, plasma T and DHT levels were measured eight times, over a period of 50 weeks, in 169 middle-aged and elderly men with symptomatic benign prostatic hyperplasia, who were otherwise healthy. The results show an excellent correlation (r = 0.849) between plasma androgen levels at first sampling and the mean of the 7 samples taken subsequently over 1 yr. Of the 22 subjects with plasma testosterone levels below the lower limit of normal (10 nmol/L) at first sampling, none had an annual plasma T value greater than 13.5 nmol/L, i.e. the annual mean was also below the normal limit or in the low normal range, whereas of the 18 subjects with a T level at first sampling above 25 nmol/L (+/- 90th percentile), none had a mean plasma T level below 19.0 nmol/L (70th percentile). Similar results were obtained for DHT. When plasma testosterone levels at first sampling were subdivided into groups, with a concentration interval of 5 nmol/L from 5 to more than 25 nmol/L, multifactor analysis of variance showed no significant difference between the values within the same group at each sampling, whereas at each sampling the groups remained highly significantly (P less than 0.001) different from each other. The same applies to DHT levels when subdivided into groups with a concentration interval of 0.5 nmol/L. It is concluded that in healthy middle-aged and elderly men, single point plasma androgen measurements reflect fairly reliably the annual mean androgen level.