Background and Objective: Prostate cancer is an endocrine-dependent tumor which is still under-investigated for physiopathology factors related to its natural history. The association of pretreatment total testosterone (TT) serum levels with prostate cancer is still a controversial topic. The objective of this study was to investigate potential associations and functional relationships of preoperative TT serum level and pathology-detected Gleason score (pGS). Materials and Methods: Pretreatment and pathological variables of 220 patients operated with radical prostatectomy were retrospectively reviewed. Age, prostate-specific antigen (PSA), percentage of positive biopsy cores (P+), biopsy Gleason score (bGS), pGS, TT and free testosterone were the continuous variables, while clinical stage (cT: cT1c, cT2/3), biopsy Gleason pattern (bGP: ≤3+3, 3+4, >3+4), pathology Gleason pattern (pGP: ≤3+3, 3+4, >3+4), pathology stage (pT: pT2, pT3a, pT3b), pathology nodal staging (pN: pN0, pN1, pNx) and surgical margin invasion by cancer (R-, R+) were the categorical variables. Statistical methods were computed for assessing associations of TT and pGS; moreover, simple and multiple linear regression analysis (SLRA and MLRA) were used for assessing functional relationships of TT and pGS. Results: High-grade tumors (pGS ≥8.0) were associated with bGS >6.0 (p < 0.0001), pGP ≥3+4 (p < 0.0001), P+ >0.31% (p = 0.006), cT2/3 (p = 0.01), TT >15.5 nmol/l (p = 0.0004) and, to a lesser extent, PSA >6.27 μg/l (p = 0.06). The odds ratio (OR) ranked as follows: 2.01 (PSA >6.27 μg/l), 2.88 (cT2/3), 3.23 (P+ >0.31%), 5.53 (TT >15.5 nmol/l) and 12.09 (pGP ≥3+4 and pGS ≥8.0). On SLRA, pGS variation was significantly predicted by bGS (p < 0.0001), P+ (p < 0.0001), PSA (p = 0.0005) and TT (p = 0.02); on MLRA, pGS variation was still significantly predicted by bGS (p < 0.0001), P+ (p = 0.04), PSA (p = 0.03) and TT (p = 0.002). When bGS, P+, PSA and TT were dichotomized to their median value, only bGS (p < 0.0001) and TT (p = 0.001) showed independence in predicting pGS variation. The best model for predicting pGS variations was by dichotomizing TT above its median (>15.5 nmol/l) because the predictive coefficient increased to 0.32, which means that patients with TT >15.5 have a significantly higher estimated risk for high-grade pGS than patients with TT ≤15.5 nmol/l (OR = 1.31). Conclusion: In a patient population undergoing radical prostatectomy, increased pretreatment serum measurements of TT are associated with and functionally related to high-grade pGS; moreover, baseline TT together with bGS and PSA are important factors for predicting pGS and assessing high-grade tumors. Baseline TT serum levels might have prognostic potential for assessing treatment response for continuous as well as intermittent androgen deprivation therapy.