We attempted to perform risk categories based on the free/total prostate-specific antigen ratio (%fPSA), prostate-specific antigen (PSA) density (PSAD, in ng ml−2), and multiparametric magnetic resonance imaging (mpMRI) step by step, with the goal of determining the best clinical diagnostic strategy to avoid unnecessary tests and prostate biopsy (PBx) in biopsy-naïve men with PSA levels ranging from 4 ng ml−1 to 10 ng ml−1. We included 439 patients who had mpMRI and PBx between August 2018 and July 2021 (West China Hospital, Chengdu, China). To detect clinically significant prostate cancer (csPCa) on PBx, receiver-operating characteristic (ROC) curves and their respective area under the curve were calculated. Based on %fPSA, PSAD, and Prostate Imaging-Reporting and Data System (PI-RADS) scores, the negative predictive value (NPV) and positive predictive value (PPV) were calculated sequentially. The optimal %fPSA threshold was determined to be 0.16, and the optimal PSAD threshold was 0.12 for %fPSA ³0.16 and 0.23 for %fPSA <0.16, respectively. When PSAD <0.12 was combined with patients with %fPSA ³0.16, the NPV of csPCa increased from 0.832 (95% confidence interval [CI]: 0.766–0.887) to 0.931 (95% CI: 0.833–0.981); the detection rate of csPCa was similar when further stratified by PI-RADS scores (P = 0.552). Combining %fPSA <0.16 with PSAD ³0.23 ng ml−2 predicted significantly more csPCa patients than those with PSAD <0.23 ng ml−2 (58.4% vs 26.7%, P < 0.001). Using PI-RADS scores 4 and 5, the PPV was 0.739 (95% CI: 0.634–0.827) when further stratified by mpMRI results. In biopsy-naïve patients with PSA level of 4–10 ng ml−1, stratification of %fPSA and PSAD combined with PI-RADS scores may be useful in the decision-making process prior to undergoing PBx.