The efficacy of vancomycin in pseudomembranous colitis was assessed in a prospective randomised controlled trial. Forty-four patients with postoperative diarrhoea were allocated to five days' treatment with either 125 mg vancomycin six-hourly or a placebo. Sixteen patients had high titres of the neutralised faecal toxin characteristic of pseudomembranous colitis; nine received vancomycin and seven placebo. At the end of treatment faecal toxins were present in one patient given vancomycin compared with five of the controls. Vancomycin caused the disappearance of Clostridum difficile from the stool in all except one patient, whereas toxicogenic strains of Cl difficile persisted in all but one of the controls. Histological evidence of psuedomembranous colitis had disappeared by the end of treatment in six out of seven patients given vancomycin compared with only one out of seven patients given vancomycin compared with only one out of five patients given placebo. In patients with faecal toxins bowel habit had returned to normal in seven of the vancomycin group compared with only one of the controls, but there was no significant difference in clinical response among patients without faecaal toxins. The results suggest that vancomycin eliminates toxin-producing Cl difficile from the colon and is associated with rapid clinical and histological improvement in patients with pseudomembranous colitis.