The optimum conditions for the detection of human immunoglobulin G (IgG), IgM, and IgA antibodies to Ureaplasma urealyticum by an enzyme-linked immunosorbent assay (ELISA) were established by using a cell lysate antigen and commercially available alkaline phosphatase conjugates. No significant cross-reactions were observed among rabbit antisera to a variety of mycoplasmas of human origin and ureaplasma antigen, thus demonstrating the specificity of the ELISA. All human sera were assayed at a 1:200 dilution. Antigen was used at 20 mug of protein/ml and conjugates were diluted 1:500. Presence of IgG antibody to U. urealyticum was significantly associated with isolation of U. urealyticum (P < 0.001) in 110 women. Seventeen acute-phase and 19 convalescent-phase sera from male nongonococcal urethritis (NGU) patients were tested for the presence of antibody by both the metabolism inhibition assay and by ELISA, with overall agreements of 82 and 95% for acute- and convalescent-phase sera, respectively. Serum antibody responses were demonstrated to selected serotypes in the metabolism inhibition test, but the response as measured by the ELISA was independent of the serotype of the antigen used. Serum antibody levels in NGU patients were significantly higher (P < 0.002) than the normal serum standard in the IgG, IgM, and IgA classes. Additionally, the magnitude of change between acute- and convalescent-phase sera was greater for NGU patients than for normal asymptomatic ureaplasma-positive male controls. A significant change in antibody levels of one or more antibody classes was detected for 12 of 18 (67%) NGU patients by ELISA. Ten of the 12 (83%) individuals had a change in the IgM class, which is suggestive of an active infectious process. The ELISA is advantageous in that it requires only a single serotype antigen, uses one serum dilution, is class specific, and allows quantitative detection of differences between acute- and convalescent-phase sera.