OBJECTIVE The circulating testosterone levels of healthy men decline with advancing age. This process is characterized by considerable inter-individual variability, the causes of which are of significant biological and clinical interest but remain poorly understood. Since sleep quantity and quality decrease with age, and experimentally-induced sleep loss in young adults results in hormonal changes similar to those that occur spontaneously in the course of aging, this study examined whether some of the variability in circulating testosterone levels of older men can be related to objective differences in their sleep. METHODS Observational study. METHODS General community and university clinical research center. METHODS Twelve healthy men ages 64 to 74 years. METHODS Three morning blood samples were pooled for the measurement of total and free testosterone. In addition to overnight laboratory polysomnography, wrist activity monitoring for 6-9 days was used to determine the amount of nighttime sleep of the participants in everyday life settings. RESULTS The main outcome measures were total sleep time and morning testosterone levels. Sleep time in the laboratory was correlated with the usual amount of nighttime sleep at home (Pearson's r = 0.842; P = 0.001). Bivariate correlation and multiple linear regression analyses revealed that the amount of nighttime sleep measured by polysomnography was an independent predictor of the morning total (Beta 0.792, P = 0.017) and free (Beta 0.741, P = 0.029) testosterone levels of the subjects. CONCLUSIONS Objectively measured differences in the amount of nighttime sleep are associated with a significant part of the variability in the morning testosterone levels of healthy older men.