We investigated whether rises in anti-double stranded DNA (anti-ds DNA) antibody levels prior to disease exacerbations of systemic lupus erythematosus (SLE) are part of a restricted immune response or merely the consequence of polyclonal B cell activation. As possible inducers of polyclonal B cell activation, we analyzed, in addition, the role of clinically apparent infections in relation to changes in levels of anti-ds DNA and disease exacerbations. We prospectively followed 72 lupus patients who were examined for disease activity and infections at least every 3 months by history and physical examination according to a protocol. Once a month, we measured levels of IgG-class antibodies to an unrelated recall antigen (tetanus toxoid), levels of IgG-class antibodies to a viral antigen (cytomegalovirus late antigens [CMV-LA]), levels of total immunoglobulins G and M, and levels of anti-ds DNA (by ELISA and Farr assay). Thirty-three exacerbations and 31 infections were observed during a follow-up period of an average of 18.5 patient months. Twenty-four out of the 27 exacerbations accompanied by a positive test for anti-ds DNA were preceded by a significant rise in anti-ds DNA: in 17 cases by ELISA and in 22 cases by Farr assay. These 24 rises in levels of anti-ds DNA prior to the exacerbation were paralleled by a significant rise in levels of IgG-class anti-tetanus antibodies in 8 cases, anti-CMV-LA antibodies in 9 cases, total IgG in 7 cases, and total IgM in 15 cases. Median rise in anti-ds DNA, as measured both by ELISA and Farr assay, exceeded the median rise in anti-tetanus antibodies, anti-CMV-LA antibodies, and total IgG and IgM (P less than 0.0001). Only one infection was recorded within a period of 3 months prior to an exacerbation, whereas infections never occurred within a period of 3 months prior to a rise in anti-ds DNA. We conclude that rises in anti-ds DNA prior to exacerbations of SLE are largely due to preferential activation of anti-ds DNA-specific B cells and not merely to polyclonal B cell hyperactivity. Clinically significant infections are not related to rises in levels of anti-ds DNA nor to the induction of exacerbations in SLE.