CIA can be viewed as a multifactorial animal model of experimentally-induced autoimmunity that is targeted to joint tissues and under multiple gene control. Thus, although induction of CIA requires immune reactivity to type II collagen, a high immune response to type II collagen is not pathognomonic of CIA, indicating that determinant specificity is of crucial importance. Also, both RT1-linked and non-RT1-linked gene directed functions are involved in the final clinical response to immunization with type II collagen. RT1-linked control is likely exerted at the level of Class II (Ia) molecules (as it is in mice) with inherent selectivity of arthritogenic vs non-arthritogenic epitopes for presentation to the immune response system; non-RT1-linked control may reflect genes controlling T-cell receptors, immunoglobulin subtypes or complement components. There is also evidence that the effects of potentially pathological anti-collagen autoimmunity may in some strains be muted or even obviated by other non-RT1 gene controlled traits that are not directly related to the immune system. These general conclusions are in close accord with those of other investigators who have carefully conducted extensive and in-depth studies of the immunogenetics of CIA in mice. CIA is obviously not an exact model of any one of the more common rheumatic diseases, such as rheumatoid arthritis or systemic lupus erythematosus. In fact, it is more closely analogous to polychondritis and some of the other sero-negative connective tissue diseases. However, CIA remains an extremely useful model in attempts to understand the genetic and environmental factors which influence a specific and definable autoimmune process--anti-collagen reactivity. In turn, autoimmunity to collagen, and to other autoantigens, is a contributing or complicating aspect of most of the diverse human rheumatic disease syndromes which have been identified to date. The characteristics of the CIA model in rats which have been discussed in this article, i.e., genetically controlled variations in incidence, severity, rate of progression and expression of clinical disease, are also characteristic of the human rheumatic disease patient population. Likewise, the probable contribution of multiple genes to these syndromes is recognized. Continued investigation of the CIA model can be expected to yield important information that can be used to better understand its human counterparts.