To identify factors that affect normalization of laboratory measures after treatment for neurosyphilis, 59 subjects with neurosyphilis underwent repeated lumbar punctures and venipunctures after completion of therapy. The median duration of follow-up was 6.9 months. Stepwise Cox regression models were used to determine the influence of clinical and laboratory features on normalization of cerebrospinal fluid (CSF), white blood cells (WBCs), CSF protein concentration, CSF Venereal Disease Research Laboratory (VDRL) reactivity, and serum rapid plasma reagin (RPR) titer. Human immunodeficiency virus (HIV)-infected subjects were 2.5 times less likely to normalize CSF-VDRL reactivity than were HIV-uninfected subjects. HIV-infected subjects with peripheral blood CD4+ T cell counts of < or =200 cells/ mu L were 3.7 times less likely to normalize CSF-VDRL reactivity than were those with CD4+ T cell counts of >200 cells/ mu L. CSF WBC count and serum RPR reactivity were more likely to normalize but CSF-VDRL reactivity was less likely to normalize with higher baseline values. Future studies should address whether more intensive therapy for neurosyphilis is warranted in HIV-infected individuals.