PURPOSE: We aimed to determine the positive and negative predictive values of multiparametric MRI (MP-MRI) for extraprostatic extension (EPE) at radical prostatectomy (RP) for different prostate cancer (PCa) risk groups. MATERIALS AND METHODS: We evaluated a cohort of 183 patients that underwent 3 Tesla (3-T) MP-MRI, including T2-weighted, diffusion-weighted MR Imaging (DWI) and dynamic contrast-enhanced (DCE) sequences, with an endorectal coil (ERC) before RP, pathological stage at RP was used as standard reference for EPE. The cohort was classified into low-, intermediate and high-risk groups according to the d'Amico criteria. We recorded prevalence of EPE at RP and determined sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) of MP-MRI for EPE in each group. Uni- and multivariable analyses were performed to identify predictors of EPE at RP. RESULTS: Overall prevalence of EPE at RP was 49.7% ranging from 24.7-77.1% between low- and high-risk categories. Overall staging accuracy of MP-MRI for EPE was 73.8 with sensitivity, specificity, PPV and NPV of 58.2 89.1 84.1% and 68.3 respectively. PPV of MP-MRI for EPE was best in the high-risk cohort with 88.8 NPV was highest in the low-risk cohort with 87.7 With an odds ratio (OR) of 10.3 MP-MRI is by far the best pre-operative predictor of EPE at RP. CONCLUSIONS: For adequate patient counselling, knowledge of predictive values of MP-MRI for EPE is of utmost importance. High NPV, important for decisions on nerve-sparing strategies at RP, is only reached in low-risk subjects.