Because a number of data studies include some controversial results about Methylenetetrahydrofolate reductase (MTHFR) polymorphisms and Down syndrome (DS), we performed a meta-analysis to determine a more precise estimation of this association. Studies were searched on PubMed, EMBASE and Lilacs-Scielo, up to April 2013, and they were eligible if they included case mothers (DSM) that have gave birth to children with DS, and controls mothers (CM) that have gave birth to healthy children without chromosomal abnormality, syndrome or malformation. The combined odds ratio with 95% confidence intervals was calculated by fixed or random effects models to assess the strength of associations. Potential sources of heterogeneity between studies were evaluated using Q test and the I(2). Publication bias was estimated using Begg's test and Egger's linear regression test. Sensitivity analyses were performed by using allelic, dominant, recessive and codominant genetic models, Hardy-Weinberg equilibrium (HWE) and ethnicity. Twenty-two studies with 2,223 DSM and 2,807 CM were included for MTHFR C677T and 15 studies with 1,601 DSM and 1,849 CM were included for MTHFR A1298C. Overall analysis suggests an association of the MTHFR C677T polymorphism with maternal risk for DS. Moreover, no association between the MTHFR A1298C polymorphism and maternal risk for DS was found. There is also evidence of higher heterogeneity, with I(2) test values ranging from 8 to 89%. No evidence of publication bias was found. Taken together, our meta-analysis implied that the T allele carriers might carry an increased maternal risk for DS.