BACKGROUND & AIMS: Celiac disease (CD) is a common chronic disorder of the small intestine, resulting from aberrant cellular responses to gluten peptides, and often remains undiagnosed. It is a complex genetic disorder, although 95% of the patients carry the risk heterodimer human leukocyte antigen (HLA)-DQ2. Genome-wide association studies on CD have identified 9 non-HLA loci that also contribute to CD risk, most of which are shared with other immune-related diseases. Our aim is to predict the genetic risk for CD using HLA and non-HLA risk alleles. METHODS: We selected 10 independent polymorphisms in 2,308 cases and 4,585 controls from Dutch, UK, and Irish populations and categorized the individuals into 3 risk groups, based on their HLA-DQ2 genotype. We used the summed number of non-HLA risk alleles per individual to analyze their cumulative effect on CD risk, adjusting for gender and population group in logistic regression analysis. We validated our findings in 436 Italian cases and 532 controls. RESULTS: CD cases carried more non-HLA risk alleles than controls: individuals carrying >= 13 risk alleles had a higher CD risk (odds ratio, 6.2; 95% confidence interval, 4.1-9.3) compared with those carrying 0-5 risk alleles. Combining HLA and non-HLA risk genotypes in one model increases sensitivity by 6.2% compared with using only HLA for identification of high-risk individuals with slight decrease in specificity. CONCLUSIONS: We can use non-HLA risk factors for CD to improve identification of high-risk individuals. Our risk model is a first step toward better diagnosis and prognosis in high-risk families and population-based screening.