Twenty-four patients with familial adenomatous polyposis who had previously undergone prophylactic colectomy and had advanced duodenal polyposis were entered into a randomized trial to assess the effect of the non-steroidal anti-inflammatory drug sulindac on duodenal and rectal polyps. Polyp size and number were assessed by videotaped duodenoscopy (and rectoscopy in 14 patients) at entry and after 6 months of treatment; the tapes were compared by two assessors who were unaware of the randomization and the shuffled chronological order of the recordings. Mucosal cell proliferation was measured by in vitro incorporation of 5-bromo-2'-deoxyuridine. Sulindac therapy was associated with a reduction in epithelial cell proliferation in the duodenum (median labelling index (LI) 15.8 versus 14.4 per cent, P = 0.003) and a trend towards duodenal polyp regression (P = 0.12). In the rectum, cell proliferation showed a marked reduction (median LI 8.5 versus 7.4 per cent, P = 0.018), and significant (P = 0.01) polyp regression was seen. Rectal polyposis was less severe than that in the duodenum and responded more dramatically. Sulindac is a possible treatment for patients in whom rectal polyps have failed to show significant regression after ileorectal anastomosis and who are unsuitable for pouch surgery; it may be useful in early duodenal polyposis or as an adjunct after duodenal clearance.