Twenty consecutive patients with the clinical suspicion of a perianal fistula were studied to define the optimal and time-efficient imaging planes for endoanal MRI in the identification and classification of perianal fistulas. The duration of each part of the MR procedure was recorded in all patients. Off-axis axial (A), coronal (C), sagittal (S) and radial (R) T2-weighted sequences were performed in all patients. Sets of images and combinations of images (A; R; CS; AR; ACS; ARCS) without patient data were reviewed in masked fashion and independently with a 2-month interval between each set. The reader was masked to the results of previous readings. The findings were compared with the surgical findings. The number of correctly identified and classified fistula and the sensitivity and specificity were determined. Twenty fistulas were present at surgery: 14 transsphincteric fistulas and six intersphincteric fistulas. Eighteen fistulas were correctly identified with the radial (R) and combined coronal sagittal (CS) sequences. In all other sequences or combinations of sequences all 20 fistulas were identified. Classification was correct with A in 16 patients, with R in 15, with CS in 15, with AR in 18, with ACS in 17 and with ARCS in 18. The sensitivity and specificity were optimal using AR or ARCS (0. 86 and 1 respectively). The optimal and most time-efficient imaging protocol for endoanal MRI of perianal fistulas thus comprises the axial and radial imaging planes.