It is recognized that testosterone (T) levels decrease in men with age, as does sexual function. We hypothesize that T supplementation in hypogonadal men with sexual dysfunction will restore certain elements of sexual function. Hypogonadal male subjects (total T < or = 300 ng/dL, n = 406, mean age 58 years) reporting one or more symptoms of low testosterone were randomized to T gel (50 mg/d and 100 mg/d), T patch, or placebo. Twenty-four-hour pharmacokinetic profiles for T were obtained. The 3 primary end points evaluated at 30 and 90 days posttreatment included a significant change in the frequency of intercourse and nighttime erections per 7-day week as well as a change in sexual desire measured on a Likert-type scale and calculated as a mean daily score. At day 30, a significant increase from baseline in sexual desire was noted for those on 100 mg/d T gel compared with those on 50 mg/d T gel, T patch, or placebo (1.2 vs 0.4, 0.7, and 0.4, respectively). A significant increase from baseline in the frequency of nighttime erections was also noted for those on 100 mg/d T gel compared with those on 50 mg/d T gel or placebo (51% of subjects in the 100 mg/d T gel group had an increase in frequency vs 30% for the 50 mg/d T gel group and 26% in the placebo group). Finally, a significant increase from baseline in the frequency of intercourse was evidenced for those on 100 mg/d T gel compared with those on T patch or placebo (39% of subjects in the 100 mg/d T gel group had an increase in frequency vs 21% for the T patch group and 24% in the placebo group). Similar results were seen for 100 mg/d T gel at day 90 for sexual desire and nighttime erections vs placebo. These data demonstrate a clear relationship between restoring serum T concentrations and improvement in certain parameters of sexual function. We propose that threshold T levels are needed in order to significantly affect improvements in sexual functioning.