Testosterone (T) substitution in hypogonadal men results in growth of the prostate gland. T effects are mediated via the androgen receptor (AR). The length of the (CAG)n polymorphism of the AR gene is negatively associated with transcriptional activity and might account for variations in prostate growth during substitution therapy. In 131 hypogonadal men aged 18-69 yr, we assessed prostate volume longitudinally by transrectal ultrasonography and determined AR (CAG)n, sex hormone levels, and anthropometric measures. Sixty-nine men with primary and 62 with secondary hypogonadism began substitution therapy with im injections of T enanthate (n = 81), transdermal T preparations (n = 19), sc injections of human chorionic gonadotropin (n = 17), or oral T undecanoate (n = 14) for 2.4 +/- 0.8 yr. Average prostate size increased from 15.8 +/- 6.1 ml to 23.0 +/- 6.8 ml. ANOVA including covariates revealed initial prostate size to be dependent on age (P < 0.001) and baseline T levels (P = 0.01) but not on number of (CAG)n (ranging from 13-30; mean, 21.4 +/- 3.5). Prostate growth per year and absolute prostate size under substituted T levels (6.1 +/- 3.3 to 21.6 +/- 10.3 nmol/liter) were strongly dependent on (CAG)n, with lower treatment effects in longer repeats (both P < 0.001). Other significant predictors were initial prostate size (negative for growth rate and positive for absolute size) and age (positive for both growth rate and absolute size). The odds ratio for men with (CAG)n less than 20, compared with those with (CAG)n of 20 or more to develop a prostate size of at least 30 ml under T substitution, was 8.7 (95% confidence interval, 3.1-24.3; P < 0.001). This observation was strongly age dependent with a more pronounced odds ratio in men older than 40 yr. This first pharmacogenetic study on androgen substitution in hypogonadal men demonstrates a marked influence of the AR gene (CAG)n polymorphism on prostate growth.