Introduction: The purpose of this study was to compareprostate-specific antigen (PSA), PSA density (PSAd), the prostateimaging-reporting and data system (PI-RADS) score, and lesiondimension (four parameters) in clinically significant prostatecancer (PCa) detection.Methods: This study included 154 patients who underwentmulti-parametric prostate magnetic resonance imaging(mpMRI) and 12 quadrant systematic prostate biopsy between01/2018 and 03/2019. Two radiologists used the PI-RADSversion 2.1 to describe the MRI findings by consensus. A Gleasonscore ≥3+4 was assessed as clinically significant PCa. Areasunder the curve (AUC) were calculated using receiver operatingcharacteristics. Youden’s index was used to determine idealcutoffs. DeLong’s test was used to evaluate statisticallysignificant differences between the four parameters.Results: The median age was 66 (±6.9) in this cohort. Themedian PSA level was 7.8 ng/dL (±18.4, 1.6-109.3), the medianPSAd was 0.146 ng/mL/cm3, and the median lesion dimensionwas 12 mm. In MRI, the number of cases with the PI-RADSscores from 1 to 5 were 34, 6, 11, 38, and 65, respectively. Interms of pathology, there was no tumor in 44 patients’ samples,while insignificant cancer and clinically significant PCa wereseen in 33 and 77, respectively. The AUC values of PSA, PSAd,PI-RADS score, and lesion dimension were 0.684, 0.731, 0.856,and 0.858, respectively. The optimal cutoffs were ≥10 ng/mL forPSA, ≥0.22 ng/mL/cm3 for PSAd, ≥4 for the PI-RADS score and≥10 mm for lesion dimension. DeLong’s tests showed that thePI-RADS score and lesion dimension were significantly superiorto PSA and PSAd. There was no significant difference betweenthe PI-RADS score and lesion dimension. Conclusion: The PI-RADS score and lesion dimension hadhigher accuracy than PSA and PSAd in clinically significantPCa detection. Lesions ≥10 mm were associated with the riskof clinically significant PCa, and this should be considered inreporting.