Background. High-risk prostate cancer (PCa) occurs in 15-25 % of newly diagnosed cases and is a life-threatening condition that requires active treatment. In recent years, the percentage of high-risk PCa has significantly increased, as well as the number of prostatectomies performed in patients with unfavorable morphologic features. However, the high-risk group criteria are not fully defined yet. According to various medical associations, a locally advanced or localized disease may have a high risk of progression. Study objective: to evaluate early and long-term results of treatment of patients with high-risk PCa depending on the high-risk group criteria. Materials and methods. The analysis includes results of radical surgical treatment of 832 patients with localized or locally advanced high-risk PCa treated in three medical institutions in St. Petersburg in the period from 2001 to 2019. Clinically high-risk group included patients with one of the following criteria: prostate specific antigen level >20 ng/ml, Gleason score >8, stage (cT); according to the last criterion two groups of patients were identified: HR-EAU (≥cT2c; n = 408) and HR-NCCN (≥cT3a; n = 282). Results. The average prostate specific antigen level was 21.09 and 26.63 ng/ml, respectively, in HR-EAU and HR-NCCN groups (p< 0.0001). The incidence of positive surgical margin, positive lymph nodes (pN+), five-year recurrence-free, cancer-specific, and overall survival did not differ significantly between the clinically high-risk groups. When evaluated according to the criteria obtained from pathomorphological examination of the removed prostate, the HR-NCCN group showed higher frequency of positive surgical margin (24.8 % vs. 19.2 %) and frequency of pN+ (22.4 % vs. 10.4 %). Analysis of long-term outcomes showed less favorable 5-year results in the HR-NCCN group (recurrence-free, cancerspecific, overall survival - 54.8, 87.0, 83.7 %) compared to the HR-EAU group (recurrence-free, cancer-specific, overall survival - 71.0, 92.1, 88.2 %) (p <0.02 for all). Conclusion. Differences in the high-risk group criteria by clinical indicators between associations do not affect early (frequency of positive surgical margin, pN+) and long-term (recurrence-free, cancer-specific, overall survival) outcomes. Pathomorphological indicators are less favorable when evaluated according to NCCN. According to our results, any of the proposed models can be used before radical prostatectomy to determine the prognosis of high-risk PCa patients. However, the NCCN morphological prognostic factors allow better prediction of outcomes and, in accordance with them, prescribe treatment that corresponds to the aggressiveness of the disease.