Forty patients with irritable bowel syndrome received an antigen-exclusion diet identical to that previously reported to be highly effective in this condition. Only 15% of the group as a whole showed evidence of food intolerance manifested by an improvement in their irritable bowel symptoms. In a further 12.5% only well-being improved and this did not seem to be related to the exclusion of any particular food. Patients whose bowel dysfunction was characterized by diarrhea responded the best (3/8) whereas those with constipation consistently failed to improve.