About 70 genetic studies have already addressed the need of biomarkers to predict the response of patients with rheumatoid arthritis (RA) to methotrexate (MTX) treatment. However, no genetic biomarker has yet been sufficiently validated. Here, we aimed to replicate a selection of 25 SNPs in the largest collection of patients up to date, which consisted of 915 patients treated with MTX. The change in disease activity (measured as ΔDAS28) from baseline was considered the primary outcome. In addition, response according to widely used criteria (EULAR) was taken as secondary outcome. We considered consistency between outcomes, P values accounting for the number of SNPs, and independence from potential confounders for interpretation of the results. Only the rs1801394 SNP in MTRR fulfilled the high association standards. Its minor allele was associated with less improvement than the major allele according to ΔDAS28 (p = 0.0016), and EULAR response (p = 0.004), with independence of sex, age, baseline DAS28, smoking, seropositivity, concomitant corticosteroid use or previous treatments. In addition, previous evidence suggests the association of this SNP with response to MTX in another autoimmune disease, juvenile idiopathic arthritis, and with high intracellular folate levels, which could contribute to poor response.