Several methods have been used to measure anti-double-stranded DNA auto-antibody (anti-dsDNA). Our aim was to determine the most efficient strategy to test anti-dsDNA in systemic lupus erythematosus (SLE). In this study, anti-dsDNA and anti-nuclear antibody (ANA) tests were requested for 644 patients. Anti-dsDNA was tested by RIA, ELISA and CLIA in all patients. The results indicated that 78 patients had a positive anti-dsDNA test according to at least one of the methods. After a 3-year follow-up period only 26 patients were diagnosed with SLE. We evaluated each method and combination of methods. Specificity and positive predictive value (PPV) increased with the number of assay methods used (p=0.002 for trend), and PPV was 100% in patients whose results were positive by all three anti-dsDNA assay methods. The proportion of anti-dsDNA-positive patients who had SLE was highest (82%; p b 0.001) among those with a homogeneous pattern of ANA staining, followed by those with a speckled pattern. In ANA positive patients, when only RIA was considered, 59% of anti-dsDNA-positive patients had SLE, but when RIA and CLIA were both considered, all patients with positive results on both tests had SLE. The combination of RIA+CLIA in patients with homogeneous and speckled ANA staining showed a similar cost and higher sensitivity than RIA alone in ANA positive patients (p b 0.001). We conclude that the most efficient strategy was to combine simultaneously two quantitative and sensitive methods but only in patients with a homogeneous or speckled pattern of ANA staining. This approach maximized specificity and PPV, and reduced costs.