OBJECTIVES: We describe studies determining the association between testosterone therapy (TTh) and mortality. MATERIALS & METHODS: We used a registry database of 737 men with adult-onset testosterone deficiency defined as presenting with low serum total testosterone (TT) levels ≤12.1 nmol/L and associated symptoms over a near 10-year follow-up. We compared associations between testosterone undecanoate (TU), cardio-metabolic risk factors and mortality using non-parametric statistics followed by separate Cox regression models to determine if any association between TU and morality was independent of age and cardio-metabolic risk factors. Finally, the association between TU and mortality was studied in men stratified by cardio-metabolic risk. RESULTS: During a median follow-up interquartile range (IQR) of 114 (84-132) months, 94 of the 737 men died. TU (ref: non-treatment) was associated with mortality; hazard ratio = 0.23, 95% confidence intervals = 0.14-0.40. Cox's regression models showed the above association to be independent of baseline age, waist circumference, hemoglobin A1c, lipids, blood pressure, smoking, and type 2 diabetes. These variables remained associated with mortality. We finally stratified the men by the high-risk baseline variables and established that the association between mortality and TU was only evident in men at higher risk. A possible explanation could lie with the "law of initial value," where greater improvements are evident following treatment in patients with worse baseline values. CONCLUSIONS: This study with long follow-up confirms that TTh is associated with lower mortality in men with adult-onset TD. This association was evident only in men with greater cardio-metabolic risk factors who demonstrated greater benefit.