In a single-blind multicenter trial, 444 patients with duodenal ulcer (DU) proven by endoscopy were randomly assigned to treatment with either ranitidine, 150 mg, or cimetidine, 400 mg, morning and evening. Clinical assessments were carried out at 2 and 4 weeks and endoscopy at 4 weeks. The patients in the 2 groups were comparable. Cumulative healing rates at 4 weeks were 78.3 p. 100 in the ranitidine group (n = 226) and 65.6 p. 100 in the cimetidine group (n = 218) (p less than 0.003). Pain at the start was of similar severity in both groups, and disappeared at the same rate under ranitidine or cimetidine: 64 p. 100 patients were painless at 1 week, 80 p. 100 at 2 weeks and 88 p. 100 at 4 weeks. Thirty-eight patients complained of mild side effects: 22 on ranitidine (2 trial withdrawals) and 16 on cimetidine (1 trial withdrawal). Multifactorial analysis (logistic model) revealed that linear ulcers had a lower healing probability than round ulcers (p less than 0.002) whatever the treatment group (cimetidine: 47 p. 100 vs 68 p. 100, ranitidine 57 p. 100 vs 80 p. 100 respectively). Smoking habits (p less than 0.057) and age less than 40 years (p = 0.056) did not significantly influence healing rates, although smokers and younger patients under cimetidine had the lowest healing rate. Thus, at the dosage used in our trial, ranitidine is more efficient for healing DU at 4 weeks than cimetidine but not for pain relief.(ABSTRACT TRUNCATED AT 250 WORDS)