The utility of incorporating biopsy information into standard risk models for prostate cancer is controversial. The percentage of positive cores (PPC) has been shown to be independently prognostic for biochemical failure (BF) but, is not usually reported. Its routine use is complicated by the practice of sometimes obtaining multiple cores from a single prostate region. Alternatively, the percentage positive core length (PPCL) and the then derived overall percentage positive biopsy tissue (PPT) have also been shown to be independently prognostic for outcome. In this study PPC and PPT were investigated, with the emphasis on regional results, as determinants of BF. From 5/89 to 8/01, 1,206 men with clinical stage T1C-3N0M0 prostate cancer received 3D conformal RT alone with a median dose of 76 Gy (65–82). Of those, 563 men had four or more core biopsies (90%≥6). The median patient age and follow-up were 67 yr (44–83) and 44 mo (24–116). The median pretreatment PSA was 7 ng/ml (0.7–73). The proportion of men with Gleason score <7 and pretreatment PSA ≤10 was 71 and 69%. The PPC was defined as the number of positive regions divided by the total number of regions. The PPT was defined as the percentage of positive tissue in the entire biopsy specimen, which was calculated by two methods: (1) using the average PPCL per region to give the APPT and (2) using the maximum PPCL per region to give the MPPT. The step-wise Cox proportional hazards model was used for multivariate analysis of BF using the ASTRO definition. Recursive partitioning analysis (RPA) was accomplished using STREE software. The PPC was not independently prognostic when tested in a model with dose (continuous, p = 0.0001), Gleason score (<7 vs. ≥7, p < 0.0001), PSA (continuous, p = 0.006), and T-stage (T1-2 vs. T3, p = 0.008). When tested separately, both the TPPT (continuous, p = 0.0012) and MPPT (continuous, p = 0.0009) were independently prognostic for BF and supplanted T-stage in models with dose (p = 0.0002 for both), Gleason score (p = 0.0002 for both), and PSA (p = 0.03 for both). When tested together, the MPPT (p = 0.0009) was significant of the two in a model with dose (p = 0.0002), Gleason score (p = 0.0002), and PSA (p = 0.03). RPA (Figure) was performed with the following covariates (omitting T-stage): MTTP (≤5, 5–10, 10–20, 20–25, 25–50, 50–75, >75%); Gleason score (<7, ≥7); PSA (≤10, 10–20, >20 ng/ml); and dose (≤72, 72–74, 74–78, >78 Gy). Differences in 5-yr Kaplan-Meier BF estimates by RPA class were statistically significant (p < 0.0001, log-rank). The MPPT is a strong, independent predictor of BF and eliminates the value of T-stage in risk stratification. The MPPT when paired with PSA identifies high grade, low volume disease that is adequately treated when ≥78 Gy is used. Similarly, MPPT identifies men with low grade, high volume disease who benefit from dose escalation.