OBJECTIVE: Infection after cerebrospinal fluid (CSF) shunts or ventriculostomies is a common complication associated with significant morbidity and mortality. Polymerase chain reaction (PCR) is a powerful molecular technique that allows rapid and precise amplification of bacterial deoxyribonucleic acid (DNA) and has proven a powerful tool in the detection of a wide variety of clinically important infectious diseases. We analyzed specimens of CSF derived from ventriculoperitoneal shunts or external ventricular drains by using both conventional cultures and PCR and report herein our preliminary results. METHODS: We selected 86 CSF samples from adult patients who underwent either shunt tap or routine surveillance cultures of their ventriculostomy. These specimens were chosen from a larger group of 300 specimens that were routinely collected (many serially) in our clinical practice. They were chosen because clinical suspicion of infection was increased because of either patient signs and symptoms (fever, stiff neck, lethargy, worsening neurological examination) or preliminary laboratory analysis of CSF data (increased white blood cell count, increased protein level, decreased glucose). We considered this subgroup optimal to efficiently initiate our investigation of the correlation of PCR and culture results. CSF was increased by using standard culture techniques and by using PCR. Samples of CSF that were to undergo PCR had DNA extracted, purified, and amplified for 16S rRNA using primers 16S-Forward and 16S-Reverse of conserved sequence regions of all bacteria. DNA was PCR-amplified for 30 cycles. One microliter of the first PCR product was subjected to nested PCR using primers specific for gram-positive and gram-negative bacteria. Samples were also subjected to PCR amplification for specific detection of Propionibacterium acnes, Staphylococcus aureus, and methicillin-resistant Staphylococcus aureus using specific primers for 16S rRNA Propionibacterium, nuclease gene of Staphylococcus, and Mec gene of methicillin-resistant Staphylococcus aureus. RESULTS: For 18 of 86 specimens (21%), both the culture and PCR were positive. For 30 of 86 specimens (35%), both the PCR and culture results were negative. For 42 of 86 specimens (49%), cultures were negative and PCR was positive. There were no positive culture results with negative PCR results. Most negative culture/positive PCR cases occurred after prolonged intravenous antibiotics. Of the 56 PCR-positive specimens, 30 were positive for Propionibacterium acnes, whereas 40 were positive for Staphylococcus aureus. Of the Staphylococcus aureus-positive specimens, two were positive for methicillin resistant-Staphylococcus aureus. Among the 56 PCR-positive specimens, 30 were positive for both Propionibacterium acnes and Staphylococcus aureus; gram-negative organisms were not detected by any method in these specimens. CONCLUSION: These preliminary data suggest that PCR is a highly sensitive, rapid, and potentially promising modality for the detection and treatment of CSF shunt ventriculostomy infection.