INTRODUCTION: The treatment of rectum cancer depends on the tumour stage, and until 2005 treatment included preoperative radiation therapy for the T3 and T4 cancer stages. An exact preoperative assessment of the cancer stage is therefore essential. In Denmark rectal Magnetic Resonance Imaging (MRI) is used as a standard procedure in preoperative evaluation, sometimes supplemented by transrectal ultrasound (TRUS). The purpose of this study was to determine the accuracy of preoperative MRI in tumour stage evaluation in order to correctly select the patients who will benefit from preoperative radiation therapy. MATERIAL AND METHODS: The MRI reports from 173 patients (98 male, 75 female, mean age 71 years) who underwent surgery for rectum cancer at Hvidovre Hospital, Copenhagen during the 2002-2005-period were evaluated. The T-stage of the MRI report was compared to the histological T-stage of the resected tumour. RESULTS: The overall accuracy of T-staging was 58% (n = 100) of which 41% T2 tumours (n = 18), 78% T3 tumours (n = 78) and 33% T4 tumours (n = 4) were correctly staged. In all, 29% of cancers were overstaged (n = 50) (100% of T1 tumours, 59% of T2 tumours, 7% of T3 tumours). A total of 13% of the cancers were understaged (15% of T3 tumours, 67% of T4 tumours). The selection of patients for preoperative radiation therapy had a sensitivity and specificity of 83% and 48%, respectively. CONCLUSION: The overall accuracy of 58% indicates that MR imaging in the early learning phases was not an optimal method for the preoperative T-staging of rectal cancer. In particular, the low specificity of MRI in selecting the patients who will benefit from preoperative radiation can result in overtreatment and increased morbidity.