OBJECTIVE More than 95% of celiac patients share the major histocompatibility complex II class human leukocyte antigen (HLA) DQ2 or DQ8 haplotype; patients negative for both types are unlikely to suffer from celiac disease. Our aim was to investigate whether HLA-DQ2 and -DQ8 typing is helpful when diagnosis is uncertain because of the absence of unequivocal small bowel villous atrophy. METHODS HLA-DQ2 and -DQ8 typing was carried out in 59 patients evincing nondiagnostic small bowel mucosal lesion or positive celiac serology, and in 17 patients maintaining a gluten-free diet without biopsy-proven celiac disease. HLA findings were compared to small bowel mucosal morphology; intraepithelial lymphocytes; and serum endomysial (EmA), reticulin, tissue transglutaminase (anti-tTG) and gliadin antibodies. RESULTS Of the 59 patients evincing only minor small bowel mucosal changes or positive celiac disease serology, 22 (37%) were negative for DQ2 and DQ8. All EmA-positive patients had celiac-type HLA, but antireticulin antibody, anti-tTg, and antigliadin antibody were also present in HLA-DQ2- and -DQ8-negative individuals. Eleven of 17 patients (65%) observing a gluten-free diet before small bowel biopsy did not share celiac-type HLA. None of the 17 had apparent villous atrophy. Serum EmA and anti-tTG were negative in all. HLA-DQ typing is less expensive than follow-up biopsy in the exclusion of celiac disease. CONCLUSIONS HLA-DQ2 and -DQ8 determination is useful in exclusion, probably lifelong, of celiac disease in individuals with an equivocal small bowel histological finding. The low specificity of this test must, however, be borne in mind.