To investigated the efficiency and accuracy of mpMRI using PI-RADS version 2 in the diagnosis of clinically significant prostate cancer. From 2010 to 2013, patients referred for prostate mpMRI underwent axial T2WI, diffusion‐weighted imaging (b=0, 800 s/mm2) and corresponding apparent‐diffusion coefficient maps, dynamic contrast‐enhanced MRI. The transrectal ultrasound (TRUS) -guided biopsies were performed within 3 months after MRI scanning. Those with a Gleason score ≥ 7 (including 3+4 with prominent but not predominant Gleason 4 component) according to the pathologic results were defined as having clinically significant cancer. Two radiologists performed the image review using PI-RADS version 2 (1 highly unlikely to be present, 2 unlikely to be present, 3 intermediate, 4 likely to be present, 5 highly likely to be present). The consistency of interobserver was evaluated by kappa test. The statistical tests were done with SPSS 18.0.Four hundred and forty-seven patients were recruited (mean age 68.7±9.0 years, PSA 15.4±14.0 ng/ml) in this study. Clinically significant cancer by biopsy was present in 195 (43.6%) patients. The diagnostic concordance of the 2 radiologists with PI-RADS was moderate (kappa=0.488). The diagnosis efficiency of clinically significant prostate cancer by 2 radiologists using PI-RADS version 2 five-score system were (scores 1=7, score 2=196, score 3=42, score 4=103, scores 5=99) and (score 1=7, score 2=171, score 3=56, score4=109, score 5=104), respectively; the receiver operating characteristic (ROC), diagnostic sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV) were 0.884 vs.0.912, 83.6%vs.89.7%, 84.5%vs.84.9%, 99.0%vs.99.1%, 21.3%vs.30.3%, respectively. The diagnostic efficiency for clinically significant cancer of mpMRI using PI-RADS version 2 shows good accuracy using PI-RADS. However the consistency of interobserver should be improved in the future.