Bi-parametric prostate MR (bp-MR) is a valuable tool for detection and characterization of prostate cancer (PCa). Recent studies suggested that PSA-density (PSA-D) in combination with multi-parametric prostate MR as well as bp-MR may achieve a higher diagnostic accuracy than either alone. We aimed to evaluate the diagnostic performance of bp-MR, PSA-D and their combination in biopsy-naïve patients.We retrospectively analyzed 334 consecutive patients who underwent prostate MR on a 3T scanner. Only patients (n = 114) who underwent TRUS-biopsy within 30 days following MR with no previous prostate biopsies were considered. Our protocol included T2-weighted and DWI sequences. A Likert score based on PI-RADS v2 was used for bp-MR evaluation. Lesions were graded histopathologically using the ISUP score. We assessed three scenarios: detection of lesions independently of ISUP score (ISUP ≥ 1), detection of both intermediate and clinically significant lesions (ISUP ≥ 2) and detection of clinically significant lesions alone (ISUP ≥ 3). Predictive value of bp-MR and PSA-D was evaluated by ROC curves and logistic regression analysis. A p value < 0.05 was considered statistically significant.In all evaluated scenarios, bp-MR showed a significantly higher predictive power (AUC = 0.87-0.95) compared to the performance of PSA-D (AUC = 0.73-0.79), while their combination (AUC = 0.91-0.95) showed no statistically significant improvement compared to bp-MR alone.Our results confirm that bp-MR is a powerful tool in detection of clinically significant PCa. Contrary to findings in the recent literature, PSA-D does not appear to significantly improve its diagnostic performance.