Recently, a treponema-specific immunoglobulin G (IgG) enzyme immunoassay (EIA), the CAPTIA Syphilis-G (Trinity Biotech, Jamestown, N.Y.), has become available as a diagnostic test for syphilis. A total of 89 stored sera previously tested by the fluorescent treponemal antibody absorption (FTA-ABS) IgG assay were evaluated by the CAPTIA EIA. The FTA-ABS IgG procedure was performed by technologists unblinded to results of rapid plasmid reagin (RPR) testing of the same specimens. Borderline CAPTIA-positive samples (antibody indices of 0.650 and 0.900) were retested; if the second analysis produced an index of >0.900, the sample was considered positive. Thirteen of 89 (15%) samples had discrepant results. Compared to the FTA-ABS assay, the CAPTIA EIA had a sensitivity and specificity and positive and negative predictive values of 70.7, 97.9, 96.7, and 79.7%, respectively. In another analysis, discrepancies between results were resolved by repeated FTA-ABS testing (technologists were blinded to previous RPR results) and patient chart reviews. Seven CAPTIA-negative samples which were previously interpreted (unblinded) as minimally reactive by the FTA method were subsequently interpreted (blinded) as nonreactive. One other discrepant sample (CAPTIA negative and FTA-ABS positive [at an intensity of 3+], unblinded) was FTA negative with repeated testing (blinded). For the five remaining discrepant samples, chart reviews indicated that one patient (CAPTIA negative and FTA-ABS positive [minimally reactive], blinded) had possible syphilis. These five samples were also evaluated and found to be negative by another treponema-specific test, the Treponema pallidum microhemagglutination assay. Therefore, after repeated testing and chart reviews, 2 of the 89 (2%) samples had discrepant results; the adjusted sensitivity, specificity, and positive and negative predictive values were 96.7, 98.3, 96.7, and 98.3%, respectively. This study demonstrates that the CAPTIA IgG EIA is a reliable method for syphilis testing and that personnel performing tests which require subjective interpretation, like the FTA-ABS test, may be biased by RPR test results.