The measurement of antibodies to DNA in SLE requires the use of double-stranded DNA (dsDNA), demonstrably free of single-stranded regions. Such dsDNA preparations can, however, contain other structural components. In this study DNA preparations with defined structure, both secondary (single- and double-stranded and random base-paired) and tertiary (superhelical and open circular), were used in the Farr assay to measure the DNA binding of sera from patients with SLE and related connective-tissue diseases. The presence of true single-stranded DNA regions in denatured DNA, native DNA, and dsDNA containing single-stranded regions increased the DNA binding measured in all sera. DsDNA, whether intact or containing small regions of random base-pairing, was bound by sera from the majority of patients with SLE but not by non-SLE sera. Superhelical dsDNA from bacteriophage PM2 was bound by SLE sera to a greater extent than linear dsDNA was. Inhibition experiments suggested that this difference in binding to DNA according to tertiary, as opposed to secondary, structure is because there are fewer available binding sites on superhelical dsDNA. DNA binding, as measured by the Farr assay, can thus be influenced by both secondary and tertiary DNA structure. Using superhelical DNA, advantage can be taken of the dsDNA form plus tertiary structure to enhance DNA binding of SLE sera beyond the levels achieved using linear dsDNA.