Anti-tumor necrosis factor (TNF) therapy heals Crohn's fistulas clinically, but the rate, extent, and duration to achieve fistula track healing are unknown. We sought to monitor deep healing, as indicated by magnetic resonance imaging (MRI), and to use this to determine treatment duration. Clinical and MRI fistula healing (at 6, 12, and 18 months), Crohn's Disease Activity Index (CDAI), Perianal Crohn's Disease Activity Index (PDAI), and the Inflammatory Bowel Disease Questionnaire were prospectively assessed. Thirty-four consecutive patients with perineal fistulas were treated with infliximab (19), adalimumab (7; all infliximab failures) and thalidomide (8). Median follow-up was 110 weeks (range, 74-161). Baseline MRI: 38% >or=2 tracks, 21% anolabial/rectovaginal. At latest follow-up, clinical fistula 'response' and 'closure' were seen in 50 and 46% of antibody-treated patients, respectively. All patients stopped thalidomide early due to side effects. Of 26 antibody-treated patients, at 6 (n=25), 12 (n=25), and 18 (n=20) months, respectively, MRI showed complete healing (20, 28, and 30%, respectively), improvement (68, 72, and 65%), no change (12, 0, and 0%) or worsening (0, 0, and 5%). MRI healing at 6 months (n=5) persisted at 12 and 18 months, including in two patients who stopped treatment at 6 months. Fistula history length and complexity did not influence the outcome. The only surgical intervention was seton insertion in one patient. The PDAI and CDAI scores decreased, and quality of life improved significantly at last follow-up. MRI fistula resolution was variable and slower than clinical healing. Prolonged treatment is often required for internal track resolution. Preliminary data suggest once MRI healing has occurred fistulas remain healed, while remaining on, or stopping anti-TNFalpha therapy. The use of a second antibody is clinically valuable.