The aim of this study was to document the appearances of fistula-in-ano on magnetic resonance imaging (MRI) and to prospectively evaluate the accuracy of MRI in the pre-operative assessment of anal fistulae. Patients with a clinical diagnosis of fistula-in-ano and awaiting surgery (n = 35) were examined with MRI. The fistulous tracks with their secondary extensions and abscesses are readily seen as low signal on T1-weighted images and high signal areas on STIR images. In order to determine the accuracy of the MRI interpretations, an experienced coloproctologist operated on all 35 patients without the knowledge of the scan interpretations and the findings at surgery were compared with the MRI scan interpretations. Concordance rates between MRI and operative findings were 86% for presence and course of the primary track, 91% for the presence and site of secondary extensions or abscesses and 97% for the presence of horse-shoeing. Although operative findings by an experienced coloproctologist were taken as the gold standard, we have shown that in 9% of our study group, failure of healing was related to pathology missed at surgery which had been documented on pre-operative MRI. It is probable, therefore, that the accuracy of MRI is higher than the figures quoted above. MRI is advocated as the imaging method of choice in the assessment of anal fistulae and its use may lead to a reduction in the recurrence rate due to inaccurate surgical assessment.