The International Society of Urological Pathology (ISUP) 2014 consensus meeting recommended a novel grade grouping for prostate cancer that included dividing Gleason score (GS) 7 into grade groups 2 (GS 3+4) and 3 (GS 4+3). This division of GS 7, essentially determined by the percent of Gleason pattern (GP) 4 (< or >50%), raises the question of whether a more exact quantification of the percent GP 4 within GS 7 will yield additional prognostic information. Modifications were also made by ISUP regarding the definition of GP 4, now including 4 main architectural types: cribriform, glomeruloid, poorly formed, and fused glands. This study was conducted to analyze the prognostic significance of the percent GP 4 and main architectural types of GP 4 according to the 2014 ISUP grading criteria in radical prostatectomies (RPs). The cohort included 585 RP cases of GS 6 (40.2%), 3+4 (49.0%), and 4+3 (10.8%) prostate cancers. Significantly different 5-year biochemical recurrence (BCR)-free survival rates were observed among GS 6 (99%, 95% confidence interval [CI]: 97%-100%), 3+4 (81%, 95% CI: 76%-86%), and 4+3 (60%, 95% CI: 45%-71%) cancers (P<0.01). Dividing the GP 4 percent into quartiles showed a 5-year BCR-free survival of 84% (95% CI: 78%-89%) for 1% to 20%, 74% (95% CI: 62%-83%) for 21% to 50%, 66% (95% CI: 50%-78%) for 51% to 70%, and 32% (95% CI: 9%-59%) for >70% (P<0.001). Among the GP 4 architectures, cribriform was the most prevalent (43.7%), and combination of architectures with cribriform present was more frequently observed in GS 4+3 (60.3%). Glomeruloid was mostly (67.1%) seen combined with other GP 4 architectures. Unlike the other GP 4 architectures, glomeruloid as the sole GP 4 was observed only as a secondary pattern (ie, 3+4). Among patients with GS 7 cancer, the presence of cribriform architecture was associated with decreased 5-year BCR-free survival when compared with GS 7 cancers without this architecture (68% vs. 85%, P<0.01), whereas the presence of glomeruloid architecture was associated with improved 5-year BCR-free survival when compared with GS 7 cancers without this architecture (87% vs. 75%, P=0.01). However, GS 7 disease having only the glomeruloid architecture had significantly lower 5-year BCR-free survival than GS 6 cancers (86% vs. 99%, P<0.01). Multivariable Cox proportional hazards regression model for factors associated with BCR among GS 7 cancers identified age (hazard ratio [HR] 0.95, P<0.01), preoperative prostate-specific antigen (HR 1.07, P<0.01), positive surgical margin (HR 2.70, P<0.01), percent of GP 4 (21% to 50% [HR 2.21], 51% to 70% [HR 2.59], >70% [HR 6.57], all P<0.01), presence of cribriform glands (HR 1.78, P=0.02), and presence of glomeruloid glands (HR 0.43, P=0.03) as independent predictors. In conclusion, our study shows that increments in percent of GP 4 correlate with increased risk for BCR supporting the ISUP recommendation of recording the percent of GP 4 in GS 7 prostate cancers at RP. However, additional larger studies are needed to establish the optimal interval for reporting percent GP 4 in GS 7 cancers. Among the GP 4 architectures, cribriform independently predicts BCR, whereas glomeruloid reduces the risk of BCR. Distinction should be made between cribriform and glomeruloid architectures, despite glomeruloid being considered as an early stage of cribriform, as cribriform confers a higher risk for poorer outcome.