This study was conducted to verify the reliability of brush cytology in detecting Helicobacter pylori in an unselected group of patients with duodenal ulcer (DU) and nonulcer dyspepsia (NUD). Endoscopy was performed on 416 consecutive patients: group A, 94 with active DU; group B, 176 patients with DU after omeprazole (n = 78), ranitidine (n = 43), or triple anti-H. pylori therapy (n = 55); and group C, 146 patients with NUD. During endoscopy, the gastric mucosa was brushed and two biopsy samples from the antrum and body were obtained for histology. In 65 patients, culture of the brush-collected materials also was performed as was that from of biopsy samples. The overall frequency of H. pylori presence detected by brush cytology was significantly higher compared with that of histology (p < 0.001), particularly in group A (p < 0.05), group C (p < 0.05), and in patients with DU after omeprazole treatment (p < 0.01), but not in patients with DU after ranitidine or anti-H. pylori treatment. The overall frequency of H. pylori-positive cultures from the brush-collected material was higher compared with cultures from the biopsy samples (38.5% vs. 24.6%), particularly in the NUD group (32.6% vs. 16.3%). Brush cytology is more sensitive than histology, besides being faster and cheaper, for the assessment of H. pylori infection, particularly when the density of the bacteria is low.