We compared immunoblotting and indirect enzyme-linked immunosorbent assay (ELISA) using different antigen preparations to test for antibody to Borrelia burgdorferi in patients with early Lyme disease. With immunoblotting, 16 (53%) of 30 patients had positive tests in acute-phase sera and 25 (83%) had them in convalescent-phase sera. Among 64 controls, false-positive results were obtained in only three individuals with syphilis and in one hospitalized patient with renal allograft rejection. Among the patients with Lyme disease, both IgM and IgG antibodies most commonly bound to the 41-kilodalton (kDa) flagellar antigen, but many patients had binding to other components, particularly those of 25, 55, 58, or 66 kDa, and the order of their appearance was variable. Compared with indirect ELISA (using sonicated whole spirochetes or a flagellin-enriched fraction as the antigen preparation), more patients with Lyme disease had positive tests by immunoblotting, and fewer control subjects had false-positive results. Our results indicate that immunoblotting is superior to indirect ELISA for diagnosing early Lyme disease.