BackgroundSeveral case–control studies have been conducted to clarify the association between the tumor necrosis factor alpha (TNF-α) -G308A polymorphism and risk of osteoarthritis (OA); however, the results are inconsistent. This meta-analysis was performed to clarify this issue using all the available evidence.MethodsEligible articles were retrieved by searching PubMed, Web of science and Google scholar. The strength of the association between the TNF-α -G308A polymorphism and risk of OA was assessed by odds ratios (ORs) with the corresponding 95% confidence interval (CI) for each study.ResultsSeven studies were included in the meta-analysis, which included 983 OA cases and 1355 controls. The pooled analysis based on all included studies showed a significantly increased OA risk in the recessive genetic model analysis (OR = 11.08, 95% CI = 4.75-25.86, p < 0.001) and in the A allele vs. G allele analysis (OR = 2.30, 95% CI = 1.08-4.90). However, there was no statistical difference in the dominant genetic model analysis (OR = 2.45, 95% CI = 0.95-6.27, p = 0.06). Furthermore, we found that OA patients had a higher frequency of the AA genotype (OR = 10.49, 95% CI = 4.47-24.61) and GA genotype (OR = 1.78, 95% CI = 1.03-3.08) compared with the control population.ConclusionOur results suggested that the TNF-α -G308A polymorphism were associated with an increased risk of OA.Electronic supplementary materialThe online version of this article (doi:10.1186/1471-2474-15-373) contains supplementary material, which is available to authorized users.