OBJECTIVE: To assess the genetic effect on the occurrence of rheumatoid arthritis (RA) associated autoantibodies. METHODS: A co-twin control study of 27 monozygotic (MZ) and 51 dizygotic same sexed (DZss) RA discordant twins. Outcome measures: The probandwise concordance rate of anti-keratin antibodies (AKA), anti-cyclic citrullinated peptide antibodies (ACPA), IgA- and IgM rheumatoid factor (IgA-RF and IgM-RF). The odds ratio for these autoantibodies based on both conditional and unconditional logistic regression adjusting for the two major genetic risk factors as well as smoking. RESULTS: The probandwise concordance rates (95% CI) of ACPA, AKA, IgM-RF and IgA-RF were 78.6 (55.4-92.4), 16.7 (0.6-58.4), 30.0 (7.3-60.6), 42.1 (14.5-71.1) in MZ twins and 25.0 (10.3-44.4), 0.0 (0.0-27.7), 10.5 (1.4-31.5) and 22.2 (6.8-45.0) in DZss twins. In twin pairs discordant for both RA and autoantibodies the OR of ACPA, AKA, IgM-RF and IgA-RF was 5 (0.5-236.5) 9 (1.3-394.5) 272 (3.5-593.2) and 10 (1.4-434.0) in MZ twin pairs and 17 (4.4-146.1) 20 (3.2-828.0) 33 (5.5-1342.4) and 577 (7.4-1149.2) in DZss twin pairs. In multiple logistic regression analysis on ACPA, the MZ/DZ OR was 21.1 (3.3-213.5) when adjusting for age, sex, ever smoking, PTPN22 1858 T-allele, Shared Epitope (SE) and SE-smoking interaction. CONCLUSION: There is a genetic contribution to ACPA generation independent of both SE and PTPN22 1858 T-allele. Environmental factors may trigger the expression of IgA-RF, ACPA and AKA in healthy persons who are predisposed to RA.