Celiac disease (CD) is a rare example of multifactorial disorder in which a genetic test is of great clinical relevance, as the disease rarely develops in the absence of specific HLA alleles. We typed DR-DQ genes in 437 Italian children with celiac disease, 834 first-degree relatives, and 551 controls. Of patients, 91% carried DQ2 and/or DQ8 heterodimers, 6% only had beta2 chain, 2% was alpha5 positive, and four were DQ2/DQ8/beta2/alpha5 negative. Only the presence of alpha5 resulted negatively associated to disease (p = 2 x 10(-4)), whereas we confirmed the effect of the beta half of DQ2 dimer on CD predisposition (p = 4 x 10(-12)). Considering 1:100 disease prevalence, we obtained a risk gradient ranging from 1:7 for DQ2 and DQ8 individuals down to 1:2518 for subjects lacking all predisposing factors. The DQB1*02 and DQB1*0302 concurrence (p = 9 x 10(-4)), besides the DQB1*02/*02 homozygosity, had an additional role in disease genetic determination. The CD prevalence rose to 17.6% in sisters, 10.8% in brothers, and 3.4% in parents. In the three groups, the subjects carrying high-risk HLA molecules were 57%, 71%, and 58%; among them, 29%, 15%, and 6% respectively had CD. Those siblings and parents with no susceptible factors were not affected. These findings indicate the impact of the HLA test for CD in clinical practice.