Currently, antibiotic efficacy in acute bacterial meningitis is evaluated in several stages. First, animals are used to assess antibiotic penetration into cerebrospinal fluid (CSF) in the absence and presence of an inflammatory stimulus. Second, concentrations of drug are correlated to in vitro killing studies. Third, clinical evaluations compare the new drugs to currently available antimicrobic--but drug failures nevertheless occur, e.g. the experiences with cephalosporins and aminoglycosides. We propose that brain tissue levels of antibiotics are an additional parameter to be monitored. Drugs that penetrate the brain substance should yield higher ventricular concentrations than drugs that penetrate the choroid plexus alone. A protective benefit may also be afforded to brain tissue per se. Experience with chloramphenicol, which penetrates the blood-brain barrier, supports these concepts; so also, do the failures with cephalosporins and aminoglycosides which, despite high CSF concentrations of these agents, afford evidence that currently monitored parameters are inadequate predictors of therapeutic efficacy.