Celiac disease (CD) is an immune disease triggered by the cereal antigen gliadin, resulting in villous atrophy in the small intestine. Susceptibility to the development of CD is strongly influenced by genes in the major histocompatibility complex, in particular alleles of the DQ genes in the class II region. However recent evidence has suggested that the major histocompatibility complex (MHC) class III region may be linked to celiac disease independently of the class II region. Among the genes located in this area is TNF-alpha, which encodes the cytokine tumor necrosis factor-alpha which has a broad range of pro-inflammatory, immunomodulatory and catabolic activities. Therefore, aberrant expression of TNF-alpha could be important in the pathogenesis of MHC-associated immune disorders. A TNF-alpha variant with a polymorphism in its promoter region has been described and designated TNF2. TNF2 has been associated with a variety of MHC-linked diseases, including systemic lupus erythematosus, dermatitis herpetiformis and insulin-dependent diabetes mellitus (IDDM), as well as parasitic infections. TNF2 has previously been shown to be associated with the MHC haplotype HLA A1-B8-DR3-DQ2, which confers susceptibility to CD. We have analyzed the distribution of TNF2 alleles in a group of celiac patients (n = 52) compared to controls (n = 52) in an effort to evaluate its role, if any, in susceptibility to the condition. TNF2 has a frequency of 0.5000 (SE +/- 0.0490) in CD, compared to 0.1635 (+/- 0.0362) in a control sample (p < 10(-6)). Of 52 patients, 44 carried one or more TNF2 alleles. Analysis indicates that the distribution of TNF2 is best explained by assuming 100% allelic association between it and HLA-DQB1*0201 (frequency = 0.7791 +/- 0.0447). However, the number of TNF2 heterozygotes significantly exceeds expectations and measurements of linkage disequilibrium confirm that allelic associations spanning the DQ and TNF regions are strongly maintained in CD. Taken together, these results indicate that TNF2 may have a role in the pathogenesis of CD; however, since it is not an independent association, the possibility that TNF2 constitutes a passive component of the CD haplotype cannot be excluded.