We assessed the intraluminal pH and mucosal prostaglandin E2 (PGE2) concentrations in patients (nonresponders) whose gastric ulcer did not heal after an initial 2 months and subsequent 3 months of treatment with one of five H2-receptor antagonists (H2RAs). The efficacy of misoprostol, a PGE1 analogue, or AG-1749, a proton-pump inhibitor, on ulcer healing in nonresponders was also tested. The percentage of time (24-h measurement) during which the pH was 3 or more was higher in the basal condition (no H2RA) in nonresponders than in good responders with healing in the first 2 months. H2RAs increased this percentage satisfactorily in nonresponders as well as in good responders. The mean mucosal PGE2 concentration in the ulcer rim was 26 ng/g of tissue in nonresponders, much lower than in good responders. Treatment with misoprostol combined with an H2RA resulted in 60% success for ulcer healing within the next 2 months in nonresponders, without affecting the intraluminal pH measured during treatment with an H2RA alone. AG-1749 raised the percentage of time with high pH to 99% from the 68% obtained with an H2RA and cured 88% of this type of ulcer. These results suggest that extremely low PGE2 levels in gastric mucosa are related to the effectiveness of the H2RA. In such a condition, the small amount of gastric acid that is still secreted even during treatment with an H2RA may interrupt ulcer healing. Therefore, treatment with a prostaglandin analogue combined with an H2RA or treatment with a proton-pump inhibitor may be useful for ulcer healing in nonresponders.