AIM: To determine whether the pattern of involvement of the gastric mucosa in lymphocytic gastritis is predictive of the presence or absence of duodenal pathology. METHODS: 50 cases (M:F, 26:24; median age 57 years) diagnosed as lymphocytic gastritis between 1986 and 1998 with concurrent duodenal (D2) biopsies were identified from a computer search of the pathology records and validated by counting gastric intraepithelial lymphocytes. Gastric and duodenal intraepithelial lymphocyte counts were performed on haematoxylin and eosin (H&E) and anti-CD3 stained sections. D2 biopsies were assessed for villous atrophy and chronic inflammatory cell infiltration by subjective grading, and gastritis was classified and graded according to the updated Sydney system. A case was designated corpus predominant when the corpus chronic inflammation grade exceeded that of the antrum. If it was less, then the case was antrum predominant, and if they were equal it was diffuse (pan-) gastritis. The ratio between the corpus and antral intraepithelial lymphocyte count in individual patients was calculated. RESULTS: Of 50 cases of lymphocytic gastritis, 21 were classified as corpus predominant. With one exception (a case of mild villous atrophy), all were accompanied by normal duodenal morphology. Cases with a corpus predominant gastritis had median duodenal intraepithelial lymphocyte counts of 19 (H&E) and 14.1 (CD3), whereas 29 subjects with an antrum predominant or diffuse gastritis had median counts of 39.9 (H&E) and 37.9 (CD3). Fifteen of these 29 cases (52%) showed villous atrophy; all were graded as moderate or severe. Patients with any degree of villous atrophy had a mean corpus/antrum intraepithelial lymphocyte ratio (H&E) of 0.59 (representing antral predominance), while those with normal duodenal morphology had a ratio of 2.39 (p < 0.0001). CONCLUSIONS: The pattern of involvement of gastric mucosa in lymphocytic gastritis is closely related to the associated duodenal pathology. Those with the corpus predominant form are unlikely to have duodenal pathology, while those with an antral predominant or diffuse form should have distal duodenal biopsies taken to exclude villous atrophy.