To ascertain the specificity of IgA and IgG antigliadin (IgA-AGA, IgG-AGA), IgA-antireticulin (R1-ARA), and antiendomysial (AEA) antibodies for the diagnosis of celiac disease, we evaluated 133 type I diabetic children aged 1.4-28.4 years (mean 14.1 +/- 6.6), with diabetes from onset to 20.5 years. Fifty-three patients were considered at onset and 49 of these also during follow-up. IgA-AGA and IgG-AGA were determined by enzyme-linked immunosorbent assay (ELISA), R1-ARA and AEA by indirect immunofluorescence. IgA-AGA were positive in 20 of 133 (15%), IgG-AGA were positive in seven of 133 (5.26%), while R1-ARA and AEA were positive in three patients. At the onset of disease we found elevated IgA-AGA in 17 of 53 (32%) patients, IgG-AGA in four (7.55%) patients, three of them with IgA-AGA as well; R1-ARA and AEA were present in three (5.66%) patients, all with high IgA-AGA levels. During 1-10 year follow-up IgA-AGA decreased to within the normal range in 13 patients, with elevated IgA-AGA at onset but without R1-ARA and AEA; in four patients with high IgA-AGA at onset, IgA-AGA remained constantly elevated as did R1-ARA and AEA in three of them; and two patients, without IgA-AGA, R1-ARA, and AEA at onset, became positive for all three antibodies. Intestinal biopsy confirmed a diagnosis of celiac disease in five of these with IgA-AGA, R1-ARA, and AEA, but not in one patient with persistent IgA-AGA but no AEA and R1-ARA, suggesting that R1-ARA and AEA are more reliable markers for the screening of celiac disease in type I diabetic patients.