It has been hypothesized that polymorphisms of Tumor Necrosis Factor (TNF)-alpha gene affect the risk of developing Alzheimer's disease (AD). However, results of different studies are often inconsistent. Our aim was to investigate by meta-analysis the association of the common polymorphisms comprehensively defining the genetic variability of the TNF-alpha gene with AD risk. Hence, the results being stated are of a meta-analysis across studies, and that this meta-analysis does not invalidate the results of the individual studies previously performed. Seventeen studies that investigated the association between 5 TNF-alpha polymorphisms (-850, -308, -863, -238, and -1031) and AD were retrieved and analyzed. The model-free approach was applied to meta-analyze these case-control genetic association studies. Available data suggested a significant association between -850 polymorphism and AD risk (TT vs. TC+CC: pooled odds ratio [OR], 1.61; 95% confidence interval [CI], 1.08-2.29; p=0.02) with no evidence of between-study heterogeneity (chi(2), p>0.1). Subgroup analysis suggested that the possession of T allele significantly increased the risk of AD associated with carriage of the apolipoprotein E epsilon 4 allele in Caucasian Australians and Northern Europeans (TT+TC vs. CC: OR, 1.95; 95% CI, 1.45-2.62; p=0.00001; p>0.1; chi(2) for heterogeneity, p>0.1). No significant difference in genotype distribution of -308 polymorphism in AD was found, with a high degree of between-study heterogeneity. For the -863 and -1031 polymorphisms we did not find an association with AD, but significant between-study heterogeneity discouraged genotype data pooling. Only four studies investigated the -238 variant and the results were not significant. Current findings support an association between -850 C>T polymorphism and the risk of developing AD; hence, they strengthen the suggestion of a potential role for anti-TNF therapy to maintain physiologic levels of TNF-alpha.