To determine the prevalence and the clinical and serological findings of neurosyphilis in HIV-infected patients, Treponema pallidum hemagglutination (TPHA) tests, CD4+ lymphocyte counts and determination of rapid plasma reagin (RPR) titers were performed in 972 HIV-infected patients over a period of 3.5 years. Patients were scored according to the Centers for Disease Control's classification for HIV infection. Reactive serum syphilis tests and positive cerebrospinal fluid (CSF)-Venereal Disease Research Laboratory (VDRL) tests, with or without clinical symptoms, were used as the criteria for diagnosis of neurosyphilis. The TPHA test was positive in 31 patients, representing 3.1% of all HIV-infected patients included in the study. Of these, 13 were intravenous drug addicts, 14 were homosexuals and 4 were heterosexuals. Diagnosis of syphilis was concurrent with HIV infection in 19 patients, prior to HIV infection in 6 patients and after HIV infection in 6 patients. CSF examinations were performed in 28 of the 31 (90.3%) patients with serologically evident syphilis. Four patients had positive CSF-VDRL tests with pleocytosis (23.5% of untreated syphilis patients in whom CSF was examined), three of whom reported mild headache, which was considered a doubtful manifestation of neurosyphilis. Patients with syphilis diagnosed and treated prior to diagnosis of HIV infection did not have evidence of neurosyphilis. Seven patients had pleocytosis with a negative CSF-VDRL test, without any clinical manifestations of neurosyphilis. There was no significant difference in the mean CD4+ lymphocyte count between patients with and without neurosyphilis (p = 0.5). RPR titers in neurosyphilis patients were greater than those in patients previously treated for syphilis and in those with pleocytosis only (p = 0.046 and 0.036, respectively). All neurosyphilis patients had an RPR titer > 1:8. After therapy, neurosyphilis patients had negative CSF-VDRL tests with a lower level of pleocytosis. The prevalence of neurosyphilis was 0.4% in HIV-infected patients and 23.5% in HIV-infected patients with untreated syphilis. This high prevalence of neurosyphilis warrants CSF examination in HIV-infected patients with syphilis, regardless of the stage of syphilis.