BACKGROUND Paradoxical external anal sphincter contraction during attempted defecation (anismus) is thought to be an important contributor to chronic faecal retention and encopresis in children. Biofeedback training can be used to teach children to abolish this abnormal contraction. METHODS A randomised controlled trial in medical treatment resistant and/or treatment dependent children with anismus using surface electromyographic (EMG) biofeedback training to determine whether such training produces sustained faecal continence. Up to four sessions of biofeedback training were conducted at weekly intervals for each patient. Anorectal manometry was performed before randomisation and six months later. Parents of patients completed the "child behaviour checklist" (CBCL) before randomisation and at follow up. RESULTS Sixty eight children underwent anorectal manometry and EMG. Of these, 29 had anismus (ages 4-14 years) and were randomised to either EMG biofeedback training and conventional medical treatment (BFT) (n = 14) or to conventional medical treatment alone (n = 15). All but one child were able to learn relaxation of the external anal sphincter on attempted defecation. At six months' follow up, laxative free remission had been sustained in two of 14 patients in the BFT group and in two of 15 controls (95% confidence interval (CI) on difference, -24% to 26%). Remission or improvement occurred in four of 14 patients in the BFT group and six of 15 controls (95% CI on difference, -46% to 23%). Of subjects available for repeat anorectal manometry and EMG at six months, six of 13 in the BFT group still demonstrated anismus v 11 of 13 controls (95% CI on difference, -75% to -1%). Of the four patients in full remission at six months, only one (in the BFT group) did not exhibit anismus. Rectal hyposensitivity was not associated with remission or improvement in either of the groups. Mean CBCL total behaviour problem scores were not significantly different between the BFT and control groups, but there was a significant improvement in CBCL school scale scores in the BFT group, and this improvement was significantly greater than that seen in the control group. CONCLUSIONS The result of this study, together with those reported in other controlled trials, argues against using biofeedback training in children with encopresis.