Basal gastric acid output was analyzed prospectively in 110 patients with endoscopically documented duodenal ulcer disease to determine the frequency of gastric acid hypersecretion in patients with bleeding versus nonbleeding ulcers. Thirty-eight patients with stigmata of an actively or recently bleeding duodenal ulcer had a mean basal output of 12.6 +/- 8.9 meq/hr. In comparison, 72 patients with nonbleeding duodenal ulcers (and no history of prior bleeding) had a significantly lower mean basal acid output of 8.7 +/- 7.5 meq/hr (P < 0.05). Twenty-four of the 38 patients (63%) with bleeding and 28 of the 72 (39%) with nonbleeding duodenal ulcers had gastric acid hypersecretion, defined as a basal acid output of greater than 10.0 meq/hr. There was a statistically significant association between bleeding duodenal ulcer and acid hypersecretion (P = 0.01). All 110 patients were treated with standard doses of H2-receptor antagonists for eight weeks. In that time, 87 patients healed and 23 patients (14 with prior bleeding and nine with nonbleeding duodenal ulcers) remained unhealed. Significantly more patients who had bled had nonhealing duodenal ulcers (P = 0.004). Irrespective of bleeding history, all 23 patients with nonhealing duodenal ulcers at eight weeks had basal acid outputs of greater than 10.0 meq/hr (range 10.1-49.1 meq/hr). These 23 patients with nonhealing duodenal ulcers were treated with increased doses of ranitidine (mean 690 mg/day, range 600-1200 mg/day) for up to eight additional weeks. All were observed to have complete endoscopic healing documented within that period.(ABSTRACT TRUNCATED AT 250 WORDS)