Background: The prostatic apex is most popular location of positive surgical margin (PSM) at radical prostatectomy (RP) and the frequency of apex is reported to be about 20-40% of all positive cases. Prostatic apex is also reported to lack a well-defined capsule. Some recommendation suggest that reporting require location of a PSM, as well as length of PSM, though there are also reports that these classifications is inadequate in some patients. In addition, there is a lack of tumor grade at PSM in these recommendations. This study aims to investigate the impact of location, number, length, and tumor grade at PSM on biochemical recurrence (BCR) after RP. Methods: We retrospectively evaluated 1013 patients with prostate cancer who underwent radical prostatectomy without neoadjuvant or adjuvant therapy at the hospitals that the authors were affiliated with between 2005 and 2013. All prostatectomy specimen slides were reviewed by a single genitourinary pathologist according to ISUP 2019 criteria. PSM were subcategorized according to the length of PSM (≤ 5mm or > 5mm), Gleason pattern (GP3 vs. GP4 or 5) at PSM, the number (single or multiple), and the location (apex-only or others). Results: The median patient age was 67 (range, 45–80) years. The median initial PSA was 6.8 ng/ml (0.4–82 ng/ml). The median follow-up period was 69 (0.7–135) months. Pathological T stage was in 73% of pT2 (n=733), 22% of pT3a (n=229), and 4% of pT3b (n=41). PSM was found in 377 cases. Length of margin is total and 5mm in median. PSM was found at apex-only in 187 cases (50%), and others showed other single location or multiple PSM. Single PSM is found in 293 cases (78%). GP3 at PSM was found in 130 cases, GP4 in 214, and GP5 in 33. GP3 at PSM tended to be more common in apex-only PSM. GP5, extra prostatic extension (EPE), seminal vesicle invasion, IDC-P, and PSM were significant in all the patients to predict BCR. Among the patients with PSM, multivariate analysis showed that length and GP at PSM were significantly strong factors and were stronger than location and number of PSM. The prognosis in pT2 patients with PSM were significantly subcategorized into 3 groups according to the length and GP at PSM (P<0.001). Conclusions: We proposed the recurrence risk classification based on the length and the GP at PSM to predict BCR after radical prostatectomy. It is useful for prognosis prediction and high-risk group should be considered for adjuvant radiation and/or hormonal therapy in postoperative care.