BACKGROUND & AIMS Susceptibility to celiac disease (CD) is related to HLA-DQ2 and DQ8 alleles and the heterodimers they encode. The objective of this study was to stratify risk for CD on the basis of HLA-DQ genotype. METHODS DNA from 10,191 subjects who are at risk for CD was analyzed for HLA-DQ haplotypes. Individuals with CD were identified as those who tested positive for anti-endomysial immunoglobulin A (EMA+) in an immunofluorescence assay. RESULTS Samples homozygous for DQ2.5 (HLA-DQA1 05-DQB1 02) or DQ2.2/DQ2.5 (HLA-DQA1 05-DQB1 02 and HLA-DQA1 0201-DQB1 02) comprised 5.38% of the total; 28.28% of these were EMA+ (95% confidence interval [CI], 24.55-32.26). Of the samples that were DQ2.5 heterozygous (HLA-DQA1 05-DQB1 02); 9.09% were EMA+ (95% CI, 7.82-10.51). Among samples in which HLA-DQ8 (HLA-DQA1 03-DQB1 0302) was detected, 8.42% of homozygotes (95% CI, 3.71-15.92) and 2.11% of heterozygotes (95% CI, 1.43-3.00) were EMA+. Samples with DQ2.2/DQ8 or DQ2.5/DQ8 comprised 5.08% of the total, and 11.78% of these were EMA+ (95% CI, 9.13-14.87). HLA-DQ2 and HLA-DQ8 were absent in 4283 samples (42.03% of the total); 0.16% of these samples were EMA+ (95% CI, 0.07-0.34). CONCLUSIONS High-resolution, sequence-specific oligonucleotide probe typing with 35 DQA1-specific and 37 DQB1-specific probes of DNA from more than 10,000 subjects was used to stratify risk of CD in an at-risk U.S. population. DQ2 homozygosity (DQ2.5/DQ2.2+2.5) increased risk for CD, estimated by the rate of EMA positivity, compared with the entire sample population and other DQ genotypes. These data suggest a quantitative relationship between the type/proportion of DQ heterodimers and the risk of CD and identify potential immunotherapeutic targets.