The treatment, diagnosis and therapeutic monitoring of hematogenous Candida meningoencephalitis (HCME) are not well understood. We therefore studied the expression of (1-->3)-beta-D-glucan (beta-glucan) in cerebrospinal fluid (CSF) and plasma in a nonneutropenic rabbit model of experimental HCME treated with micafungin and amphotericin B. Groups studied consisted of micafungin (0.5 to 32 mg/kg) and amphotericin B (1 mg/kg) treatment groups and the untreated controls (UC). Despite well-established infection in the cerebrum, cerebellum, choroid, vitreous humor (10(2) to 10(3) CFU/ml), spinal cord, and meninges (10 to 10(2) CFU/g), only 8.1% of UC CSF cultures were positive. By comparison, all 25 UC CSF samples tested for beta-glucan were positive (755 to 7,750 pg/ml) (P < 0.001). The therapeutic response in CNS tissue was site dependent, with significant decreases of the fungal burden in the cerebrum and cerebellum starting at 8 mg/kg, in the meninges at 2 mg/kg, and in the vitreous humor at 4 mg/kg. A dosage of 24 mg/kg was required to achieve a significant effect in the spinal cord and choroid. Clearance of Candida albicans from blood cultures was not predictive of eradication of organisms from the CNS; conversely, beta-glucan levels in CSF were predictive of the therapeutic response. A significant decrease of beta-glucan concentrations in CSF, in comparison to that for UC, started at 0.5 mg/kg (P < 0.001). Levels of plasma beta-glucan were lower than levels in simultaneously obtained CSF (P < 0.05). CSF beta-glucan levels correlated in a dose-dependent pattern with therapeutic responses and with Candida infection in cerebral tissue (r = 0.842). Micafungin demonstrated dose-dependent and site-dependent activity against HCME. CSF beta-glucan may be a useful biomarker for detection and monitoring of therapeutic response in HCME.