We evaluated a prototype polymerase chain reaction (PCR)-based assay for Chlamydia trachomatis developed by Roche Molecular Systems to detect endocervical infection in women. Of 587 endocervical samples obtained from women attending the Harborview Medical Center sexually transmitted diseases clinic, 58 (10%) were positive for C. trachomatis by cell culture. Compared with culture, the PCR method had a sensitivity of 88% (51 of 58) and a specificity of 99.2% (525 of 529). The positive and negative predictive values were 92.7% (51 of 55) and 98.7% (525 of 532), respectively. After resolution of discrepant results whereby true positives were considered to be either culture-positive patients (58 patients) or culture-negative patients positive upon PCR analysis using both plasmid- and major outer membrane protein-based primers (4 patients), the resolved sensitivities of the PCR and culture were 89 and 93%, respectively. We subsequently performed a second analysis of 362 women, comparing the proposed commercial PCR assay from Roche Molecular Systems with chlamydia cultures. Thirty (8%) women were infected with C. trachomatis. Compared with culture, the assay had a sensitivity of 60% (18 of 30) and a specificity of 99% (328 of 332). Repeat PCR assay done 2 to 5 days later subsequently yielded positive results for 7 of 11 PCR-negative samples from culture-positive women. We conclude that the Roche Molecular Systems PCR assay provides highly specific results compared with culture in a high-risk population of women. Further study is needed, however, to more clearly define the sensitivity of the PCR assay in detecting endocervical C. trachomatis infection in women and to identify factors that may compromise sensitivity.