Biofeedback therapy has been proposed as a treatment for fecal incontinence with good, short-term results. This study was designed to assess long-term clinical results of biofeedback therapy compared with medical therapy alone and to assess manometric results in patients treated with biofeedback. Two groups of incontinent patients were studied. Group 1 consisted of 16 patients (3 males and 13 females; mean age, 59.9 years). Etiologies treated by biofeedback included descending perineum syndrome (7), postfistula or hemorroidectomy (4), and miscellaneous (5). Group 2 consisted of eight patients (two males, six females; mean age, 62.2 years). Etiologies treated with medical treatment alone (including enema and antidiarrheal therapy) included descending perineum syndrome (3), postfistula or hemorroidectomy (2), and miscellaneous (3). The incontinence score was initially 17.81 +/- 3.27 (standard deviation) in Group 1 and 17.0 +/- 2.77 in Group 2. Resting pressure of the upper and lower anal sphincter, maximum squeezing pressure, and duration of contraction were not initially different in Groups 1 and 2 but were significantly lower than in the control group of patients without incontinence (n = 12; 8 males, 4 females; mean age, 66.4 years) (P < 0.05). Follow-up duration was 30 months, with intermediate clinical score at 6 months for Group 1. After biofeedback therapy, the incontinence score at 30 months was lower in Group 1 (14.43 +/- 6.35 vs. 17.81 +/- 3.27; P < 0.035) and unchanged in Group 2 (18.0 +/- 2.72 vs. 17.0 +/- 2.77). However, in Group 1 the score at 6 months was much lower than at 30 months (6.31 +/- 7.81 vs. 14.43 +/- 6.35; P < 0.001). Only the amplitude of voluntary contraction and upper anal pressure (51.1 (range, 27-90) vs 36.7 (range, 20-80) mmHg) were significantly increased (81.5 (range, 55-120) vs. 62.1 (range, 30-90) mmHg; P < 0.05). Biofeedback improved continence at 6 months and at 30 months. However, the score at 6 months was much better, suggesting that the initial good results may deteriorate over a long time. These data suggest that it could be useful to reinitiate biofeedback therapy in some patients.