Cross-sectional studies have demonstrated a decline in testosterone and free and bioavailable testosterone with age. This occurs in a majority of older persons without an increase in luteinizing hormone (LH), suggesting that a component of the testosterone decrease is due to secondary hypogonadism. To determine whether these findings could be duplicated in a longitudinal study, we measured testosterone, LH, follicle-stimulating hormone (FSH), and sex hormone-binding globulin (SHBG) levels in 77 men participating in the New Mexico Aging Process Study who had sera available in 1980 or 1981 and two or more serial samples in 1982, 1984, 1989, and/or 1994. Thirty-nine subjects had samples available from both 1980 and 1994. The age at entry into the study ranged from 61 to 87 years. Testosterone levels decreased over the 15 years of the study. In persons who were alive for the duration of the study, testosterone levels were significantly lower 5 years before termination of the study (P < .05). Testosterone levels did not differ at entry into the study among those who died and those who were alive at the end of the study period. Eight of 77 subjects (10%) had LH levels above the normal range at some time during the study. In contrast, 43% of subjects had elevated FSH levels. Both LH and FSH increased significantly with age. SHBG levels were measured in 1980 and 1994 and increased significantly with age (P < .0001). LH and FSH were highly correlated with one another, but neither correlated with testosterone. This study demonstrated a longitudinal decline in testosterone and an increase in LH and FSH in older men. The average rate of decrement in testosterone concentration was 110 ng/dL every decade.