To evaluate diagnostic performance of endorectal magnetic resonance (eMR) for diagnosing local recurrence of prostate cancer (PC) in patients with previous radical prostatectomy (RP) and to assess whether contrast-enhanced (CE)-eMR improved diagnostic accuracy in comparison to unenhanced study. Unenhanced eMR data of 72 male patients (mean of total PSA: 1.23 +/- 1.3 ng/ml) with previous RP were interpreted retrospectively and classified either as normal or suspicious for local recurrence. All eMR examinations were re-evaluated also on CE-eMR 4 months after the first reading. Images were acquired on a 1.5-T system. These data were compared to the standard of reference for local recurrence: prostatectomy bed biopsy results; choline positron emission tomography results; PSA reduction or increase after pelvic radiotherapy; PSA modification during active surveillance. Sensitivity, specificity, predictive positive value, negative predictive value and accuracy were 61.4%, 82.1%, 84.4%, 57.5% and 69.4% for unenhanced eMR and 84.1%, 89.3%, 92.5%, 78.1% and 86.1% for CE-eMR. A statistically significant difference was found between accuracy and sensitivity of the two evaluations (chi(2) = 5.33; p = 0.02 and chi(2) = 9.00; p = 0.0027). EMR had great accuracy for visualizing local recurrence of PC after RP. CE-eMR improved diagnostic performance in comparison with T2-weighted imaging alone.