Multiple sclerosis (MS) is a complex polygenic disease in which gene-environment interactions are important. A number of studies have investigated the association between tumor necrosis factor-α (TNF-α) -308 G/A polymorphism (substitution G→A, designated as TNF1 and TNF2) and MS susceptibility in different populations, but the results of individual studies have been inconsistent. Therefore, performing a systematic review and meta-analysis of the published studies is desirable. We sought to quantitatively summarize the association between TNF-α-308 G/A polymorphism and MS. The Medline and Scopus databases were searched to identify potentially relevant case-control studies published in English journals up to January 2010. A meta-analysis of these studies was performed. Summary odds ratios (ORs) and 95% confidence intervals (CIs) were calculated under fixed and random effects models. Twenty-one eligible studies, comprising 2880 patients with MS and 3579 controls, were included in the meta-analysis. The overall pooled ORs (95%CI) for TNF2 versus TNF1 and TNF2 carriers (2/2+2/1) versus non-carriers (1/1) were 1.02 (0.86-1.21) and 0.99 (0.8-1.24), respectively. In the European populations, the pooled ORs (95%CI) for TNF 2/1 versus 1/1 were 0.85 (0.73-0.98), which was statistically significant. However, the other results did not support this finding. The pooled ORs (95%CI) for TNF 2/1 versus 1/1 and TNF 2/2 versus 2/1 were not statistically significant in the overall population. In addition, the pooled ORs for TNF2/2 versus TNF2/1+1/1 and TNF2/2 versus TNF1/1 were not statistically significant. Our meta-analysis does not support the role of TNF-α -308 G/A polymorphism in developing MS.