Recently, a new radioimmunoassay--the polyethylene glycol (PEG) assay--was introduced to measure antibodies to double-stranded (ds) DNA. In this method, polyethylene glycol precipitation of formed 3H-DNA/antiDNA complexes is used instead of the ammonium sulfate precipitation used in the Farr assay. In contrast to the Farr assay, with which only high-avidity antibodies to dsDNA are detected, the PEG assay also reportedly measures anti-dsDNA of relatively low avidity. We studied whether this gain in antibody measurement results in loss of specificity for systemic lupus erythematosus. When the PEG assay was applied to a selected panel of 440 sera from patients with various well-defined autoimmune diseases and to a group of 197 normal human control sera, matched sex and age to the patients, the method was found to be fairly specific for systemic lupus erythematosus, although the sera from some patients with myasthenia gravis and some with autoimmune liver disease were also found positive. Screening of 352 additional serum specimens, sent to our laboratory for diagnostic reasons, revealed that, with the PEG assay, an extra population of relatively low-avidity antibodies to dsDNA--missed by the Farr assay--was detected. Upon clinical evaluation, we found that the patients in whom such antibodies were detected generally fulfilled a number of the preliminary criteria of the American Rheumatism Association for systemic lupus erythematosus, but that this diagnosis often was not made. We claim that the presence of low-avidity antiDNA characterizes a milder form of the disease in which patients often show only a single clinical feature of the disease. We conclude that results of the PEG assay add valuable diagnostic and clinical information to results obtained by the Farr assay.