To determine whether the primary grade (PG) of biopsy Gleason score (GS) 7 prostate cancer (CaP) was predictive for biochemical relapse-free survival (bRFS). Most of the present data regarding the PG of GS7 CaP refer to surgical specimens. Our goal was to determine whether the biopsy GS used at the time of medical decision making predicted for the biochemical outcome.We reviewed the data from 705 patients with biopsy GS7 CaP, from a prospectively maintained database, who had been treated at our institution between September 1996 and March 2005 with radical prostatectomy (n = 310), external beam radiotherapy (n = 268), or prostate radioactive seed implantation (n = 127). The bRFS rates were estimated using the Kaplan-Meier method. Cox proportional hazards regression analysis was used for univariate and multivariate analyses examining these factors in relation to bRFS: PG of biopsy GS, initial prostate-specific antigen level, clinical T stage, use of androgen deprivation, risk group (high or intermediate), and treatment modality.The 5-year bRFS rate was 78% and 71% (p = 0.0108) for biopsy GS7 PG3 CaP and biopsy GS7 PG4 CaP, respectively. Comparing PG3 and PG4 within treatment modalities, only prostate implantation patients had a significant difference in the 5-year bRFS rate, 88% vs. 76%, respectively (p = 0.0231). On multivariate analysis, the PG of biopsy GS remained an independent predictor of bRFS, with PG3 having better bRFS than PG4 (relative risk, 0.655; 95% confidence interval, 0.472-0.909; p = 0.0113).Biopsy GS7 PG4 CaP carries a worse bRFS than biopsy GS7 PG3 CaP.