In a comparison of commercial ligase chain reaction (LCR; Abbott) and polymerase chain reaction (PCR; Roche) assays, measuring plasmid genes of Chlamydia trachomatis, some specimens were found to be negative by either or both assays but positive in traditional culture or antigen detection tests. Of 767 women, 35 were found to be infected by cervical or urine testing. Twenty three specimens from 16 women may have contained inhibitors in six cervical swabs (CS) and 15 first void urines (FVU). By performing dilution and 'spiking' experiments on five FVU, inhibitors of PCR, LCR or both, which disappeared by dilution, were demonstrated. Confirmatory assays were used which amplified segments of the major outer membrane gene by PCR or LCR. When comparisons of assays were made on a single specimen type, the sensitivities of the amplification assays, compared to an expanded reference standard, were as follows: on CS, PCR was 93.8% (30/32) and LCR was 96.9% (31/32); on FVU, PCR was 76.6% (23/30) and LCR was 93.3% (28/30). When a combined calculation was made to determine the ability of the assays to detect patients infected in the cervix or urethra by testing FVU, the sensitivities dropped to 71.4% (25/35) for PCR and 80.0% (28/35) for LCR: CS sensitivity was 88.6% (31/35) for both amplified tests. There were two CS and five FVU false-positives by PCR which reduced to one CS and three FVU in the combined analysis. There were no false-positives by LCR. Inhibitors and low levels of chlamydial plasmid nucleic acids may have contributed to lower than expected sensitivities, suggesting a possible need for internal positive controls, especially for PCR, when testing urine. More studies with multiple sampling and more than one amplification assay are needed to confirm these findings and to identify and remove inhibitors of amplification assays.