The Alzheimer's disease “inflammation hypothesis” has emerged only recently, suggesting the risk of developing AD might be influenced by variants of genes encoding for inflammatory mediators. In order to investigate in this direction, genomic DNA from 194 Italian AD cases and 454 healthy controls matched by gender and ethnicity was analyzed for the Receptor for Advanced Glycation End products (RAGE, HLA class III-centromere portion) -374 and -429 SNPs and for the Tumor Necrosis Factor-alpha (TNF-α, HLA class III-telomere portion) -857, -308 and -238 SNPs by RFLP and Real Time PCR. Our data show statistically significant deviations between AD patients and healthy controls concerning RAGE -374 SNP genotype (TT: p=0.0084) and allele (T, A: p=0.0081) frequencies; TNF-α -308 SNP AA genotype (p=0.0433) and TNF-α -238 SNP genotype (GG: p=0.0138) and allele (G, A: p=0.0151) frequencies. Furthermore, significant differences between the study groups and regarding RAGE TC (p=0.05) and AC (p=0.009) haplotypes are present, while TNF-α haplotype reconstruction point out a statistically significant difference between patients and controls regarding AGG haplotype (p=0.002). Finally, from the combination of the individually significant SNPs of the two genes (RAGE -374, TNF-α -238 and -308) we performed an HLA class III haplotype reconstruction finding significant differences between AD subjects and controls regarding the TAG (p=0.019) and TGA (p=0.008) haplotypes. The implication of these haplotypes with the disease points to a possible involvement of entire HLA class III region in AD susceptibility.