OBJECTIVE A prospective study was designed to compare the potentials of digital rectal examination (DRE), transrectal ultrasound (TRUS), and magnetic resonance imaging (MRI) using integrated endorectal and pelvic phased-array coils for preoperative estimation of tumor volume and local extent of prostate cancer. METHODS Evaluation of 20 consecutive patients undergoing radical retropubic prostatectomy included DRE, TRUS with a 7.5-MHz transducer, and MRI on a 1.5-tesla GE Signa system. Step sections (5 mm) of the entire specimen were performed, and tumor volume and percentage of gland involved were calculated. RESULTS DRE, TRUS, and endorectal and pelvic phased-array MRI showed 50, 75, and 95% of the cancers, respectively. There was a linear correlation on MRI between predicted tumor volume and pathological tumor volume (r = 0.82, p < 0.0001), but not between predicted volume on DRE or TRUS and real volume. The accuracy for detecting extracapsular penetration was 60% for DRE and TRUS and 79% for MRI. The accuracy for detecting seminal vesicle invasion was 60% for DRE, 66 for TRUS, and 89% for MRI. The negative predictive value for extracapsular and seminal vesicle extension was highest for MRI (85 and 93%, respectively). The accuracy for tumor location in the apex of the prostate was 30% for DRE, 47 for TRUS, and 89% for MRI. CONCLUSIONS MRI with integrated endorectal and pelvic phased-array coils satisfactorily predicted tumor volume and tumor extent preoperatively. Multicoil MRI can assist in decision making as it is valuable in the definition of patients that may benefit from surgery and can be of help for evaluating the risk of a positive margin, especially in the apical resection.