We compared a fluorescein-conjugated monoclonal antibody (FA) direct specimen test (MicroTrak; Syva Co., Palo Alto, Calif.) with culture (TC) in McCoy cells (vials, with blind passage and iodine staining of inclusions) for diagnosis of Chlamydia trachomatis infection in the cervix. Duplicate specimens were collected from 1,230 women, but for 262 of these subjects, both results were unavailable (150 FA smears were inadequate, indicating a need for clinical training in specimen collection), leaving 968 comparisons. Prevalence of chlamydiae by culture was 13% (126/968). Compared with TC results, the sensitivity of FA was 70% (88/126) and the specificity was 94% (795/842). There was a 91% agreement (883/968). The predictive value of a positive FA test was 65% (88/135), and that of a negative FA was 95% (795/833). We reexamined 38 smears for which paired results were discrepant, and the reread would have changed the result in only 5 of these. TC is less than 100% sensitive and some FA-positive, TC-negative specimens represent positive specimens not detected by TC. Unfortunately, it is not possible to identify which results in this group are truly false-positive. Clearly, the FA procedure has a performance profile which would make it a useful tool in screening high-risk populations (particularly when TC is not available) but it is less suited to screening low-risk populations, for which false-positive results are more important. The greater utility of the FA procedure in a venereal disease clinic was confirmed by testing 172 evaluable specimen pairs, of which 34 (20%) were Chlamydia isolate positive. The FA sensitivity was 76% (26/34) and specificity was 96% (133/138), giving a predictive value of 84% (26/31) for a positive test.