In order to improve the diagnosis of Listeria meningitis or meningoencephalitis, especially in patients who have received antibiotics before their cerebrospinal fluid (CSF) has been examined, two assays for the detection of Listeria monocytogenes based on the polymerase chain reaction (PCR) were evaluated. After a standard PCR, the amplified DNA was detected either by a second round of PCR with internal primers followed by gel electrophoresis and ethidium bromide staining (nested PCR) or by dot blot hybridization to an internal digoxigenin-labeled probe (PCR-dot blot). For PCR, two sets of primers within the invasion-associated protein gene (iap gene) were chosen. They allowed for the highly specific detection of all L. monocytogenes reference strains tested (serotypes 1/2a, 1/2b, 1/2c, 3a, 3b, 3c, 4a, 4b, 4c, 4d, and 7). These primers did not detect amplification products from various other gram-positive or gram-negative bacterial DNAs or human DNA. The sensitivities of both assays were assessed on sterile CSF samples that were artificially seeded with serial dilutions of L. monocytogenes serotype 4b cells. By both methods the limit of detection was less than 10 cells in the initial reaction. Since the nested PCR is more prone to contamination because of manipulation of the amplified products, a standard PCR assay followed by dot blot hybridization was applied to 52 CSF samples in a retrospective study. Of 28 CSF samples which were sterile or positive for bacteria other than Listeria species, 24 were PCR negative. In contrast, from 17 patients with culture-proven Listeria meningitis, 14 of 17 initial CSF samples were PCR positive, as were 3 of 7 culture-negative followup CSF samples taken after patients received antibiotics. These results support the usefulness of this approach in the diagnosis of Listeria meningitis, in particular, when antibiotic administration precedes culture of CSF.