Ideally a diagnosis of infection of the central nervous system (CNS) is made by culture of the etiologic pathogen, but Borrelia burgdorferi, the causative agent of Lyme neuroborreliosis (LNB), is rarely cultured from the cerebrospinal fluid (CSF). PCR and measurement of specific antibody in the CSF also have their limitations. The role of available assays for LNB has not been studied carefully in a comparative investigation. There is a need to assess the reliability of assays and to increase the ability to document active infection in the CNS. The recent development of the nonhuman primate (NHP) model of LNB allowed us to address this need in a faithful model of human LNB. In this study we compared the abilities of PCR and culture to detect the presence of spirochetes in the CSF and brain tissue of infected NHPs and related these measures of infection to the development of anti-B. burgdorferi antibody. We also tested a bioassay, the mouse infectivity test (MIT), in this model. Fourteen of 16 CSFs from four NHPs were positive by at least one of these techniques. Detection of spirochetes in the CSF by PCR, the MIT, and culture was inversely related to the concomitant presence of anti-B. burgdorferi antibody intrathecally. The performance of any particular test was associated with the strength of the host immune response. In early CNS infection, when anti-B. burgdorferi antibody had not yet appeared, or in immunocompromised hosts, the MIT compared favorably to culture and PCR for infected NHPs; antibody in the CSF was the most useful assay for immunocompetent NHPs.