Antibodies to mammalian dsDNA have, for decades, been linked to systemic lupus erythematosus (SLE) and particularly to its most serious complication, lupus nephritis. This canonical view derives from studies on its strong association with disease. The dogma was particularly settled when the antibody was included in the classification criteria for SLE that developed during the 1970s, most prominently in the 1982 American College of Rheumatology (ACR), and recently in The Systemic Lupus International Collaborating Clinics (SLICC) classification criteria. There are several problems to be discussed before the anti-dsDNA antibody can be accepted without further distinction as a criterion to classify SLE. Old and contemporary knowledge make it clear that an anti-dsDNA antibody is not a unifying term. It embraces antibodies with a wide spectrum of fine molecular specificities, antibodies that are produced transiently in context of infections and persistently in the context of true autoimmunity, and also includes anti-dsDNA antibodies that have the potential to bind chromatin (accessible DNA structures) and not (specificity for DNA structures that are embedded in chromatin and therefore unaccessible for the antibodies). This critical review summarizes this knowledge and questions whether or not an anti-dsDNA antibody, as simply that, can be used to classify SLE.