Background: A prospective study was performed to determine the most reliable MRI criteria to distinguish recurrent rectosigmoid carcinoma from benign postoperative fibrosis. Methods: Twenty-two consecutive patients who were suspected to have recurrent rectosigmoid carcinoma were examined by T2 and precontrast and contrast enhanced T1 weighted images. The prospective interpretations, the presence of high signal on T2 weighted images, the shape of the margins of a mass and the degree of contrast enhancement were correlated with histology and follow up to determine their respective accuracies, sensitivities and specificities. Results: The best criteria for recurrent tumor was the combination of high signal on T2 weighted images, round margins and > 40% contrast enhancement, which had an accuracy of 92%, sensitivity of 100% and specificity of 85%. In patients who were more than one year postoperative the specificity was 100%. Conclusions: The most reliable MRI criteria for distinguishing recurrent rectosigmoid carcinoma from benign postoperative fibrosis are the combination of the signal intensity on T2 weighed images, the shape of the margins of a mass and the presence of greater than 40% contrast enhancement.