OBJECTIVES: In celiac disease (CD) the degree of histological damage in the duodenum may vary, but there is some controversy about the coexistence of villous atrophy and normal mucosa in different biopsy sites, i.e., patchy villous atrophy. We prospectively evaluated the degree, frequency, and distribution of histological lesions among different duodenal sites as well as within each duodenal biopsy. METHODS: Over the last 4 years, in each patient with suspected CD (positive anti-transglutaminase antibodies), four to five endoscopic biopsies were taken from the duodeno-jejunal flexure/distal duodenum (D3), intermediate duodenum (D2), proximal duodenum (D1), and duodenal bulb (B). Biopsies were subjected to hematoxylin/eosin staining and immunostaining with anti-CD3 monoclonal antibodies for intraepithelial lymphocyte (IEL) count. Duodenal lesions were classified according to Marsh-Oberhuber, and CD was diagnosed according to the European Society of Pediatric Gastroenterology, Hepatology, and Nutrition criteria. RESULTS: Six hundred and eighty-six children did have CD. A degree of villous atrophy was found in 660/686 patients (96.2%), total villous atrophy was present in 550/686 (80.1%), and 320/686 (46.6%) had different lesions at different sites, but none of these patients had entirely normal biopsies. In all, 116 of 686 (16.9%) had variable lesions within the same biopsy, with grade 2+3A being the most frequent association (43%), followed by 2+3A+3B (27%) and 2+3A+3B+3C (22%). All these 116 patients also had histologically normal areas within the same biopsy, but anti-CD3 immunostaining showed that IELs were always increased in such areas. In all the cases, the severity of duodenal lesions significantly increased in an aborad manner (chi(2)=52.38 with alpha=0.01 and d.f.=12; P<0.0001). No correlation was found between type and distribution of histologic lesions and clinical presentation of CD. CONCLUSIONS: In newly diagnosed CD, some variability of histological lesions can be found, even within the same duodenal biopsy, in which areas of apparently normal mucosa with increased IEL number often exist. We also confirm our previous findings that duodenal lesions may vary among different biopsies; lesion severity has a proximal-to-distal gradient, but no patient has entirely normal duodenal biopsies. The awareness of such histological variability may help establish a correct diagnosis of CD.