First-line therapy for meningitis is often penicillin plus chloramphenicol. Penicillin-resistant Streptococcus pneumoniae (PRSP) infections are increasing world wide, but the efficacy of chloramphenicol for PRSP meningitis is unknown. We therefore prospectively assessed children with pneumococcal meningitis treated with penicillin plus chloramphenicol over 27 months to compare outcome of penicillin-susceptible S pneumoniae (PSSP) meningitis with that of PRSP meningitis. 68 children with pneumococcal meningitis who survived 24 hours were evaluated, of whom 25 had chloramphenicol-susceptible PRSP meningitis that was treated initially with chloramphenicol. 20 (80%) of these 25 children had an unsatisfactory outcome (death, serious neurological deficit, or poor clinical response). By contrast, 14 (33%) of 43 children with PSSP meningitis (treated with benzylpenicillin) had an adverse outcome (p less than 0.001). Despite similar zone sizes on antibiotic disc testing (indicating chloramphenicol susceptibility) the chloramphenicol minimum bactericidal concentrations (MBCs) of PRSP isolates were significantly higher than those of PSSP isolates. The higher chloramphenicol MBCs resulted in borderline cerebrospinal-fluid bactericidal activity in many cases of PRSP meningitis and frequent treatment failure. Current definitions of chloramphenicol susceptibility of S pneumoniae may be inappropriate for management of pneumococcal meningitis. We suggest that chloramphenicol should not be used for the management of PRSP meningitis; alternative agents, such as third-generation cephalosporins, are more appropriate.