There is a paucity of data on the relationship between testosterone and Alzheimer's disease (AD) in older men. The objective of the present study was to investigate the effects of serum total testosterone (TT), bioavailable testosterone (BT), and sex hormone binding globulin (SHBG) levels on the subsequent risk of AD in nondemented Chinese older men. This was a one-year prospective cohort study. 153 ambulatory community-living nondemented Chinese older men, aged 55 years or over, were recruited and followed for one year. Morning serum TT, BT, and SHBG levels were measured at baseline. At one-year of followup, assessment for dementia and AD were performed. AD was diagnosed by the NINCDS-ADRDA criteria for probable AD. Overall, the mean age of the subjects was 72.7 (SD 6.9). 6.5% (n = 10) developed dementia (converters), all having AD. 93.5% (n = 143) did not develop dementia (non-converters). Logistic regression analysis for independent predictors of AD showed that the baseline serum BT level, systolic blood pressure (SBP) and ApoE "4 genotype were significant independent predictors, after adjustment for age, education, BMI, fasting plasma glucose, and serum HDLC levels. The baseline serum BT level predicted a reduced risk of AD (adjusted relative risk (RR) 0.22, 95% CI: 0.07-0.69)). Baseline SBP and ApoE "4 genotype but not SHBG were independent risk factors, with RRs of 1.04 and 5.04 respectively. In conclusion, the serum level of bioavailable testosterone in late life predicts a lower risk of future AD development in older men.