Co-trimoxazole was compared with nafcillin against Staphylococcus aureus in vitro and in the therapy of experimental Staph. aureus meningitis in rabbits. Co-trimoxazole (trimethoprim:sulphamethoxazole in a 1:20 ratio) was synergistic against 22/24 strains of Staph. aureus in vitro. The MBC90 of co-trimoxazole and nafcillin were 0.156-3.12 mg/l and 0.25 mg/l, respectively, concentrations below those achievable in purulent cerebrospinal fluid. The rate of bacterial killing (Staph. aureus) by co-trimoxazole and nafcillin were similar in both broth and pooled CSF in vitro. However, the MBC increased and the rate of bactericidal activity of both agents declined when tested in CSF at a higher inoculum (10(7) cfu/ml). During continuous intravenous infusion therapy of a reproducible, uniformly fatal (if untreated) model of experimental Staph. aureus meningitis, serum concentrations of all agents closely approximated those found in humans receiving standard parenteral regimens. The mean percent penetration into CSF ([CSF]/[serum] X 100) was 2.9, 35.6 and 27.1% for nafcillin, trimethoprim and sulphamethoxazole, respectively. Although both nafcillin and co-trimoxazole therapy reduced CSF Staph. aureus concentrations significantly more rapidly (P less than 0.001) when compared to untreated controls, the bactericidal rate was modest. The CSF was rendered sterile in 0/64 animals treated with either regimen for 8 h. Nafcillin was more rapidly bactericidal in vivo (P less than 0.03) than co-trimoxazole in this model. Caution is advised in the use of co-trimoxazole for infections of the central nervous system caused by Staph. aureus.