We compared a direct fluorescent-antibody stain (DFA) and an enzyme immunoassay (EIA) with a standard cell culture technique for the detection of Chlamydia trachomatis infection in women in an urban family practice setting. We also evaluated a DFA sample in a commercial laboratory to determine the interlaboratory reliability of this test. There were 268 women in the study; the EIA provided a higher sensitivity (83 versus 50%) and a higher positive predictive value (83 versus 69%) than the DFA test and comparably high specificity (99 versus 98%). Concordance between the two laboratories on the DFA test was not high when data were adjusted for chance agreement (kappa coefficient = 0.64). DFA validity was optimal with an elementary body cutoff of greater than 5, while EIA validity was optimal at the recommended cutoff of 0.1 optical density unit. None of 11 women with negative cultures after treatment had false-positive antigen tests. False-negative results with both tests were associated with low culture inclusion counts but were not strongly associated with the presence or absence of symptoms, menses, pregnancy, or recent antibiotic use. False-positive results with EIA were seen only for three women who had a chief complaint of vaginal discharge. Although the positive predictive value of DFA could be increased in high-prevalence subpopulations, EIA was still more valid in two such groups: teenagers and prenatal patients. These results indicate that EIA might be preferable for low- or moderate-prevalence populations in primary care settings and that a falloff in DFA sensitivity could be explained by lower infection burdens in low-prevalence groups.