The evidence is now fairly conclusive that collagen-induced arthritis in rodents is mediated by antitype II collagen autoimmunity. Arthritis is probably initiated by binding of antibodies to the surface of intact articular cartilage. Many of the major manifestations of arthritis, including synovial proliferation, pannus formation, marginal erosion of bone, and destruction of cartilage, can be duplicated by injection of isolated antitype II collagen antibodies. It is not known whether delayed hypersensitivity reactions to collagen can provoke similar lesions in the absence of antibody, but circumstantial evidence suggests they do not. Also clear is that not all anticollagen antibodies are capable of inducing arthritis. The minimal requirements for arthritogenic potential are currently under investigation but probably include the ability to bind native autologous type II collagen. Also IgM antibodies alone are either ineffective or are required in relatively higher concentrations than IgG for induction of arthritis. Autoimmunity to collagen is found in many spontaneous and induced rheumatic diseases other than collagen-induced arthritis. There is at present, however, no direct evidence that this autoimmunity actually contributes to the arthritic process. Nevertheless, the human disease most often associated with collagen autoimmunity is rheumatoid arthritis. In many respects the immune reactions detected in humans with rheumatoid arthritis parallel those of arthritis in rodents. That is, responsiveness is under the control of genes within or linked to the major histocompatibility locus. High responders are limited to only a few haplotypes. Cell-mediated reactions are most vigorous in response to denatured collagen and probably have limited specificity for the type of collagen recognized. Antibodies may be separated into at least two groups, one with broad specificity for denatured collagen and a second highly specific for conformation-dependent determinants on native type II collagen. The latter antibodies are of most interest to researchers because they may be like those that induce arthritis in rodents. There is also ample evidence that antibodies are deposited in the joints of rheumatoid arthritis patients, although the specificity of these antibodies is unknown. Generally, collagen-induced arthritis is a model of antibody-initiated autoimmunity arthritis. Specifically, it is a model of type II collagen autoimmune arthritis. In consideration of its extraarticular manifestations, it may justifiably be referred to as type II collagen autoimmune disease.(ABSTRACT TRUNCATED AT 400 WORDS)