Only within the last 5 years have data on variable region sequences of autoantibodies begun to accumulate. Although it is too early to draw final conclusions, certain principles are beginning to emerge. It is clear that self-recognition by the immune system is normal. This fact is supported by the anti-idiotype network and the presence of natural autoantibody in normal sera. Furthermore, there appears to be a connection between these phenomena because the former may serve to keep the latter in check. Natural autoantibody appears to be primarily IgM polyreactive antibody of low affinity, quite different from the monospecific high-affinity IgG antibody usually associated with autoimmune disease. Natural autoantibodies also exhibit a high degree of cross-reactive idiotypes and are the products of CD5+ B cells, whereas IgG autoantibody is not. This is only a generalization, however, and it must be kept in mind that IgG autoantibody polyreactive with several autoantigens has been reported for normal sera of mice and humans. In Balb/c sera, a significant proportion of natural autoantibody is IgG, but it is apparently masked by inhibitory polyreactive IgM antibody that has affinity for IgG autoantibody, thereby exhibiting a regulatory role. A human CD5+ B-cell clone that secretes a high-affinity IgM anti-idiotype-specific antibody has been reported. Another generalization that is beginning to emerge is that natural autoantibodies are the products of unmutated germline genes, whereas autoantibody associated with autoimmune disease has highly mutated VH and VL segments. Mutations exhibited by pathogenic autoantibody increase the antibody's affinity for autoantigen and therefore represent an antigen-selected or affinity maturation, resulting in more restricted heterogeneity. In contrast to this, polyreactive natural antibodies have variable regions that are essentially identical to germline gene V segments with few amino acid changes. However, it is important to note that pathologic autoantibodies arise from the same fetal genes that give rise to natural autoantibodies and antibodies against foreign antigens. There does not appear to be a set of genes specific for autoantibodies. The key question then becomes whether or not natural autoantibodies are precursors of pathogenic autoantibodies. If not, what is the function of natural autoantibodies and how do pathogenic autoantibodies arise? The answers to these questions are not yet clear and many theories have been offered, but sufficient data are not yet available to reach conclusions. One of the problems in reaching a definitive conclusion is the ability to identify pathogenic antibody.(ABSTRACT TRUNCATED AT 400 WORDS)