These days, rheumatoid arthritis (RA) is reported to be subclassified into two subsets by anti-citrullinated peptide antibody (ACPA) positivity. Clinically, ACPA positive RA tends to develop more severe arthritis than ACPA negative RA. In addition, a lot of reported susceptibility genes to RA (ie. HLA-DRB1(*)04, PTPN22, TRAF1/C5, CTLA4) are found to be associated only with ACPA positive RA but not with ACPA negative RA. It is getting clear that HLA-DRB1(*)04, which was believed to be primarily associated with RA, is not a primary risk factor but ACPA is. Then, a hypothesis for the disease mechanism of ACPA positive RA is set as follows; citrullination possibly due to smoking, etc, provokes ACPA production in individuals who have susceptibility alleles of genes including HLA, followed by joint inflammation in autoantibody-dependent manner. The search for susceptibility genes for ACPA negative RA is slowly progressing, but only a few genes are so far reported: HLA-DRB1(*)03 for Caucasian, HLA-DRB1(*)09 for Japanese, IRF5 and STAT4. When we investigate the disease mechanisms of RA, we should manage independently the two disease subsets : ACPA positive and ACPA negative RA.