The prevalence of H. pylori infection has been reported to be very high in duodenal ulcer (DU) disease, but the precise frequency and causes of H. pylori-negative DU are not well known. In some geographical regions, however, a relatively low prevalence of the infection has been described. Our aim was to study the frequency and causes of H. pylori-negative DU and to evaluate whether empirical H. pylori eradication therapy without confirmation of the infection is justified. In all 774 consecutive patients with an endoscopic diagnosis of DU were studied prospectively. Exclusion criteria were associated diseases and previous gastric surgery. The use of NSAIDs, antibiotics (during the last month), and proton pump inhibitors (during the last month) was evaluated by means of a specific questionnaire. At endoscopy, two biopsies from both antrum and corpus were obtained in all 774 patients for histologic study (H&E stain). One sample from the antrum for rapid urease test, one sample each from the antrum and corpus for culture, and two duodenal biopsies for histologic study were also obtained in the first 307 patients. A [13C] urea breath test was carried out in the remaining 467 patients. Patients were considered infected if any of the diagnostic tests were positive and noninfected when all tests performed were negative. Age (mean +/- SD) was 46+/-12 years, 70% were males. NSAID, antibiotic, and proton pump inhibitor use was described, respectively, in 8.9%, 5.8%, and 6.3% of the cases. H. pylori infection was demonstrated, overall, in 95.3% (95% CI: 93.6-96.6%) of the patients. H. pylori prevalence increased up to 99.1% (98.1-99.6%) if patients taking NSAIDs and/or antibiotics were excluded. Among the 36 H. pylori-negative patients, 20 (55%) were taking NSAIDs, 9 (25%) were taking antibiotics, and 1 (3%) both of them. Therefore, in only 6/774 patients (0.8%) could DU disease be considered truly "idiopathic." Differences were demonstrated between H. pylori-positive and -negative patients (univariate study; chi2) with regard to NSAID intake (7% vs 58%; P < 0.0001) and previous antibiotic use (5% vs 28%; P < 0.0001). In the multivariate analysis (logistic regression), NSAID use (OR: 0.06; CI: 0.03-0.13; P < 0.001) and antibiotic use (OR: 0.23; CI: 0.09-0.59; P < 0.01) were the only variables that correlated with H. pylori infection. The most important factors associated with H. pylori-negative DU are NSAIDs and prior antibiotic use, and if these agents are excluded, the prevalence of infection in our area is as high as 99%. Therefore, in DU patients not taking NSAIDs and living in areas where previous studies have shown the prevalence of the infection in DU disease to be very close to 100%, empirical H. pylori eradication therapy without confirmation of the infection may be justified.