Using an enzyme-linked immunosorbent assay (ELISA) assay, we studied the sera of 17 patients with celiac disease and 114 control subjects for the levels of IgG and IgA antigliadin antibodies. As a group, the patients with celiac disease had significantly higher levels of antigliadin antibodies of both IgG and IgA classes (p less than or equal to 0.001). However, there was a significant overlap of values, resulting in respective sensitivities and specificities of 88% and 90% for IgG antigliadin antibodies, and 73% and 65% for IgA antigliadin antibodies. The combined use of both IgG and IgA antigliadin antibody levels produced a sensitivity of 86% and a specificity of 90%. A gluten-free diet in celiac patients did not seem to affect these results. We conclude that the antigliadin antibody ELISA assay cannot be used as a definitive diagnostic test for celiac disease. The small-bowel biopsy remains the principal diagnostic method; the ELISA assay should, at best, be considered a screening test for this disease.