Thirty-eight patients with severe antibiotic-associated postoperative diarrhoea were entered into a randomized controlled trial to compare colestipol (an ion exchange resin) in 17 patients with placebo (sherbet) in 21 patients. Clostridium difficile or its toxin was present before treatment in 12 of the colestipol group, compared with only 5 in the placebo group. Because of the low incidence of Cl. difficile or its toxin, the placebo group data from 22 patients receiving placebo in a previous trial (9 of whom had Cl. difficile or toxin) were included for comparison. Neither colestipol nor placebo had any influence on the faecal excretion of Cl. difficile or its toxin. Colestipol was clinically no better than placebo. In view of the persistent faecal excretion of Cl. difficile toxin, ion exchange resins cannot be recommended for the treatment of antibiotic-associated colitis.