Testosterone (T) levels decline with age and lower T has been associated with increased mortality in aging men. However, the associations of its metabolites, dihydrotestosterone (DHT) and estradiol (E2), with mortality are poorly defined. We assessed associations of T, DHT, and E2 with all-cause and ischemic heart disease (IHD) mortality in older men. Participants were community-dwelling men aged 70 to 89 years who were residing in Perth, Western Australia. Plasma total T, DHT, and E2 were assayed using liquid chromatography-tandem mass spectrometry in early morning samples collected in 2001 to 2004 from 3690 men. Deaths to December 2010 were ascertained by data linkage. There were 974 deaths (26.4%), including 325 of IHD. Men who died had lower baseline T (12.8±5.1 vs 13.2±4.8 nmol/L [mean±SD], P=.013), DHT (1.4±0.7 vs 1.5±0.7 nmol/L, P=.002), and E2 (71.6±29.3 vs 74.0±29.0 pmol/L, P=.022). After allowance for other risk factors, T and DHT were associated with all-cause mortality (T: quartile [Q] Q2:Q1, adjusted hazard ratio [HR]=0.82, P=.033; Q3:Q1, HR=0.78, P=.010; Q4:Q1, HR=0.86, P>.05; DHT: Q3:Q1, HR=0.76, P=.003; Q4:Q1, HR=0.84, P>.05). Higher DHT was associated with lower IHD mortality (Q3:Q1, HR=0.58, P=.002; Q4:Q1, HR=0.69, P=.026). E2 was not associated with either all-cause or IHD mortality. Optimal androgen levels are a biomarker for survival because older men with midrange levels of T and DHT had the lowest death rates from any cause, whereas those with higher DHT had lower IHD mortality. Further investigations of the biological basis for these associations including randomized trials of T supplementation are needed.