Treatment of pneumococcal meningitis has become problematic because of the emergence of penicillin- and cephalosporin-resistant strains and because of the concern that dexamethasone therapy might reduce penetration of antibiotics into the cerebrospinal fluid (CSF). We addressed these issues with our rabbit meningitis model by studying two pneumococcal isolates that were resistant to penicillin and ceftriaxone and susceptible to vancomycin and rifampin. Ceftriaxone, vancomycin, and rifampin were given alone or in combination, with or without coadministration of dexamethasone. Treatment was started 12 to 14 h after intracisternal inoculation of approximately 10(4) CFU of one of the organisms. Rifampin concentrations in serum and CSF were similar, regardless of whether dexamethasone was given, whereas those of ceftriaxone were somewhat lower at each time point in animals given dexamethasone. The penetration of vancomycin into CSF was consistently and substantially reduced with dexamethasone treatment, which resulted in a delay in CSF sterilization not observed in non-dexamethasone-treated animals. When rifampin was used with ceftriaxone for treatment of meningitis caused by the more resistant strain, bacteriologic cure occurred promptly, with or without dexamethasone therapy. In areas with high rates of occurrence of resistant pneumococcal strains, we believe initial empiric therapy of bacterial meningitis should include two antibiotics: ceftriaxone and either rifampin or vancomycin. When dexamethasone is used, the combination of ceftriaxone and rifampin is preferred for therapy.