The objective of this study was to observe the dose-response effects of testosterone (T) treatment on symptoms of sexual dysfunction and the metabolic syndrome. Two cohorts of elderly men with late-onset hypogonadism were followed over 9 months. Group 1, consisting of 28 men (mean age, 61 years; mean T level, 2.07 +/- 0.50 ng/mL), received long-acting T undecanoate (TU; 1000 mg); group 2, composed of 27 men (mean age, 60 years; mean T level, 2.24 +/- 0.41 ng/mL), received T gel (50 mg/day) for 9 months. In patients treated with T gel, plasma T levels rose from 2.24 +/- 0.41 to 2.95 +/- 0.52 (statistically significant) at 3 months, 3.49 +/- 0.89 (statistically significant) at 6 months, and 3.80 +/- 0.73 ng/mL at 9 months (T level at 6 months was compared with T level at 3 months). With TU, plasma T levels rose from 2.08 +/- 0.56 to 4.81 +/- 0.83 (statistically significant) at 3 months, 5.29 +/- 0.91 at 6 months, and 5.40 +/- 0.77 ng/mL at 9 months. With TU, the plasma T levels were statistically significantly higher than with T gel With TU, there was a greater improvement in sexual symptoms and in symptoms of the metabolic syndrome. With both treatments, changes in waist circumference correlated with changes in total, low-density, and high-density lipoprotein cholesterol. Parameters of safety were not different between the 2 treatments. T administration had a beneficial effect on sexual dysfunction and symptoms of the metabolic syndrome in elderly men. The higher plasma levels of T generated with TU than with T gel were clearly more effective, indicating that there is a T dose-effect relationship.