With a view to describing the natural history of duodenal ulcer, currently linked with powerful and widely used drugs, we retrospectively reviewed all patients followed-up at our unit between 1978 and 1989. All DU subjects with at least 5 years of clinical-endoscopic follow-up were admitted to the study. Concomitant gastric ulcer, previous gastrointestinal surgery and treatment other than with H2-blockers (where necessary) were considered exclusion criteria. Forty-four patients (35 males, 9 females, mean age 44.5) proved eligible for this study. At monthly intervals in the patients' history of ulcer disease (starting from a baseline month 0 when ulcer symptoms first appeared), we compared all cases in which the duration of disease coincided, i.e. for each month of the clinical course of the ulcer there was a certain number of patients at the same point in time; patients' histories coincided variously over a period ranging from 0 to 15 years. We considered endoscopically-diagnosed relapses (both symptomatic and asymptomatic) and hemorrhages as marker events for the activity of DU disease. A year-by-year duodenal ulcer activity rate was obtained for both treated and untreated subjects, as follows: No. of events/mean No. of patients followed-up per month x 100. Fisher's exact test and multivariate analysis (stepwise logistic regression) were used to investigate risk factors of poor outcome. We found that the years between the 1st and 5th and between the 10th and 15th from the onset of duodenal ulcer disease had higher peaks of activity, mostly in untreated subjects.(ABSTRACT TRUNCATED AT 250 WORDS)