A new enzyme immunoassay (EIA, Captia Syphilis-G) for detecting IgG antibodies against Treponema pallidum was evaluated as a screening test for syphilis. When serum samples were tested at a dilution of 1 in 20 (EIA20), the overall agreement between the IgG EIA and serological status based on the T pallidum haemagglutination assay (TPHA) and the fluorescent treponemal antibody absorption (FTA-ABS) test was 99.2% (1310/1321). The sensitivity of the EIA20 was 98.4% (60/61) and the specificity 99.3% (1251/1260). Discrimination between patients with and without treponemal infection was good: the mean EIA20 absorbance ratios (patient/mean low titre positive control results) were 0.49 for antibody negative patients, 3.30 for patients with positive Venereal Diseases Research Laboratory (VDRL) test and TPHA results, and 1.77 for patients with negative VDRL but positive TPHA results. The cut off point for excluding treponemal infection was taken as 0.9. Specimens with ratios of more than 0.9 should be confirmed by the FTA-ABS test and evaluated for specific IgM antibodies to treponemes. When serum samples were tested at a 1 in 50 dilution (EIA50) the sensitivity was lower (80.3%) but the specificity was absolute. The reduction in sensitivity correlated with low absorbance ratios in the patients who were VDRL negative and TPHA positive. The screening performance of the IgG EIA20 is thus comparable with that provided by a combination of the VDRL test and TPHA. The potential for automation makes the EIA an attractive alternative, particularly in larger centres. Alternatively, the test can be performed at a 1 in 50 dilution (EIA50), at which level it is ideally suited for confirming the treponemal status of antibodies in serum samples preselected by positive cardiolipin antigen screening test results.