Diabetes mellitus is a common chronic disease, affecting 0.5-2% worldwide. The Massachusetts Male Aging Study reported that up to 75% of men with diabetes have a lifetime risk of developing ED. Type 2 diabetes is associated with low total serum testosterone (TT) identified in several cross-sectional studies and systemic analyses. There is a lack of consensus regarding what constitutes the lowest level of testosterone within the boundaries of normality. In this retrospective study, we sought to evaluate the effect of associated co-morbidities on serum total testosterone (TT) level in men with type 2 diabetes DM, either with or without erectile dysfunction (ED). Three hundred and ninety-one patients were evaluated for erectile function using an abridged, five-item version of the International Index of Erectile Function-5. Measurements of TT, fasting lipid profile, blood sugar and glycated haemoglobin (HbA1c) were conducted. Penile hemodynamics was assessed using intracavernosal injection and penile duplex study. Hypogonadism was found in 126 cases (33.2%), and normal TT was observed in 254 (66.8%). ED was detected in 119 cases in the hypogonadal group (94.4%) as compared to 155/254 (61.0%) in eugonadal group, P = 0.0001. TT was lower in diabetic men with ED as compared to those with normal erectile function (EF), 392.4 ± 314.9 versus 524.3 ± 140.2 ng dl(-1) , respectively, P < 0.0001. After exclusion of patients with hypertension and dyslipidaemia, 185 men were evaluated, and there was no difference in the mean TT level among men with ED 490.6 ± 498.2 ng dl(-1) versus normal EF 540.6 ± 133.4 ng dl(-1) although, HbA1c remained lower in men with normal erectile function. Receiver operating characteristic (ROC) curve of TT in men without associated co-morbidities showed that EF was compromised at TT = 403.5 ng dl(-1) or less. Sensitivity of 63.3% and a specificity of 94.0% were detected. At this level, ED was found in 33/38 (86.8%) men with TT 403.5 ng dl(-1) , whereas ED was observed in 57/147 (38.8%) men with TT ≥ 403.5 ng dl(-1) (P < 0.0001). We propose a cut-off value of 403.5 ng dl(-1) of TT blood levels as an indicator for initiation of testosterone replacement therapy in diabetic men with ED. Further prospective controlled trials are recommended.