BACKGROUND: /st> Rheumatoid arthritis (RA) does not always respond to available treatments, including tumour necrosis factor (TNF) antagonists. A study was undertaken to investigate whether genetic variation within genes, encoding proteins in the p38 signalling network, contribute to the variable response to TNF antagonists. METHODS: /st> 968 UK Caucasian patients with RA receiving anti-TNF therapy (infliximab, adalimumab and etanercept) were genotyped for 38 pairwise-tagging single nucleotide polymorphisms (SNPs) spanning 12 candidate genes from the p38 network. Regression analyses were performed to test association between genotype and treatment response at 6 months using both absolute change in DAS28 (Disease Activity Score across 28 joints) and European League Against Rheumatism (EULAR) improvement criteria. Stratified analyses were performed to investigate association with individual therapies. RESULTS: /st> Twelve SNPs in seven genes were associated with improvement in DAS28 at 6 months at a nominal 0.1 significance level, jointly explaining 3% of variance in outcome in a model adjusting for other predictors. These encoded proteins both upstream (RIPK1, MKK6) and downstream (MAPKAPK2, MSK1, MSK2 and MNK1) of p38, and MIF, involved in p38 MAPK deactivation. Three SNPs (rs1258012 in MKNK1, rs4240847 in MAPKAPK2 and rs475032 in RPS6KA4 encoding MSK2) were associated with the EULAR response at the 0.1 level. SNPs in four genes showed correlation with the response to infliximab and adalimumab but not to etanercept. No SNPs had a greater impact on the response to etanercept. CONCLUSIONS: /st> More SNPs than would be expected by chance, mapping to the p38 signalling network, showed association with the anti-TNF response as a whole, and particularly with the response to infliximab and adalimumab. Validation of these findings in independent cohorts is warranted.