BACKGROUND: Previous reports, based on surgery, showed duodenal ulcer (DU) to be more common in the rice-eating areas of southern India than in the northern wheat-eating areas. AIMS: Does this difference persist? Can it be explained by risk factors other than diet? METHODS: A total of 20 053 records from patients undergoing endoscopy for dyspepsia, and 590 endoscopy patients from two northern and two southern centers in India were studied prospectively. Records were scrutinized to determine the relative incidence of DU and non-ulcer dyspepsia in wheat- and rice-eating areas. Age, sex, length of history, smoking and medication were recorded. Three antral biopsies and one from each duodenal quadrant were taken. A rapid urease test was carried out on one of the antral biopsies; the others were examined for Helicobacter pylori, gastritis, duodenitis and duodenal gastric metaplasia. RESULTS: The difference in diet-associated prevalence persisted. No differences in smoking, Helicobacter pylori infection or duodenal gastric metaplasia were found between the two regions, but all three were more common in DU than in non-ulcer dyspeptic patients from both dietary areas. CONCLUSIONS: The dietary differences between the regions remain the only factor to account for the differences in DU prevalence. A strong interrelationship between duodenal gastric metaplasia and cigarette smoking is demonstrated.