Rheumatoid arthritis (RA) is a systemic autoimmune disorder based in the synovium of peripheral joints. Rheumatoid factor (RF) and anti-citrullinated protein antibodies (ACPAs) are autoantibodies detectable in the majority of patients, with the latter being highly specific for RA. Retrospective studies utilizing preclinical serum samples have demonstrated that RF and ACPAs are detectable in the serum of RA patients months to years before disease onset. Moreover, a close association between ACPAs, smoking, and disease-predisposing HLA-DRB1 alleles has been identified, suggesting that these risk factors may converge to precipitate autoantibody-positive RA. Animal models of RA have added considerable information regarding the immune events that precede joint inflammation. These models have demonstrated that autoantibodies to ubiquitous antigens can directly precipitate chronic organ-specific inflammation centred in the joint. Furthermore, it has recently been possible to demonstrate a role for ACPAs in the animal models of RA. The major challenge currently is to develop a robust predictive model for RA onset, identifying the factors that serve to initiate, amplify, and mature the immune responses towards citrullinated autoantigens. Recent data from a high-risk population of RA family members indicate that the nature and specificity of the ACPA response in healthy individuals differs considerably from that in RA patients, and support the concept that this autoimmune response evolves over time and leads to the onset of clinically detectable synovitis. Ultimately, the availability of data from prospective studies of RA onset will allow for novel strategies that can potentially prevent disease in high-risk individuals.