Objectives: To evaluate whether preoperative magnetic resonance imaging (MRI) allows prediction of complete tumour resection in patients planned for pelvic exenteration (PE). Study design: Data of all patients treated by PE at a gynaecologic cancer centre between 6/1999 and 5/ 2005 were studied retrospectively. Preoperative MRI scans were re-analysed blindly with respect to invasion of neighbouring organs, muscular pelvic side wall, vessels and lymph nodes by an experienced team of radiologist and gynaecologic oncologist, finally also giving estimation whether complete removal of tumour would be feasible. Results: 43 patients were identified: all of them underwent PE. The histopathological investigation of the exenteration specimen demonstrated microscopic complete resection status in 20 cases (46.5%). None of the investigated parameter showed a correlation to complete resection of tumour. Sensitivity for final microscopic results relative to preoperative prediction by radiographic findings was 85%, specificity was 52%. The positive predictive value for tumour-free status was 60% (17/28), the negative 80% (12/15). Conclusions: Preoperative MRI cannot predict the surgical outcome with sufficient accuracy, hence on its own is inappropriate for prediction of complete tumour resection and final histological margin status in patients undergoing PE.