The use of tripotassium dicitrato bismuthate (TDB) has been suggested recently for cimetidine resistant duodenal ulcers. This study compares the efficacy of TDB with two different cimetidine dosages in the treatment of duodenal ulcer patients who failed to respond to an eight week therapy with H2-blockers. Fifty two patients (40 men, 12 women) were randomly allocated to one of the following three oral regimens: (1) TDB 120 mg quid, (2) cimetidine 400 mg tid, (3) cimetidine 400 mg with meals plus 800 mg at bedtime. Endoscopy was carried out after four weeks; if the ulcer had not healed patients continued with the same treatment for a further four week period when they were endoscopically reassessed. After four weeks similar percentages of ulcer healing were registered in the two cimetidine schedules (39% with 1.2 g and 44% with 2 g), whereas TDB resulted in a significantly higher healing rate (82%) compared with cimetidine 1.2 g (p = 0.01) and with cimetidine 2 g (p = 0.025). After eight weeks the cumulative percentages of healing were 65% on cimetidine 1.2 g, 75% on cimetidine 2 g, and 94% on TDB (TDB v cimetidine 1.2 p = 0.042). These results confirm previous data that resistant duodenal ulcers are more responsive to an agent which strengthens the mucosal defences than to antisecretory compounds.