We studied the effect of meningitis and the method of parenteral gentamicin administration (intramusclar injection, a 30-min intravenous infusion, or intravenous bolus administration) on achievable concentrations of drug in cerebrospinal fluid (CSF). In normal animals, only intravenous bolus administration of 2 to 8 mg/kg produced a gentamicin concentration of >0.1 rg/ml in CSF in some aninalsg. All CSF samples contained less than the limit of detection (0.1 ,ug/ml) after the intramuscular administration of 6 mg/kg. In animals with meningitis, gentamicin penetration into cisternal CSF was increased significantly after a bolus administration of 6 mg/kg (mean, 0.197 ± 0.063 ,ug/ml in normal animals versus 1.68 ± 0.38 ,ug/ml in animals with meningitis; P < 0.01). In meningitic animals that received 6 mg/kg as an intravenous bolus, lumbar CSF had the highest maximum concentration (4.25 ± 1.08 pg/ml), in comparison with ventricular CSF (3.10 ± 0.66 "g/ml). The gentamicin concentration in cisternal CSF decreased more slowly than it did in serum (elimnnation half-life, 238.70 ± 64.56 min in cisternal CSF versus 82.73 ± 2.91 min in serum), yielding a relative increase in the percentage of penetration. We conclude that maximum penetration by gentamicin into CSF occurs after intravenous bolus administration and that the maximum concentration occurs in lumbar CSF.Meningitis caused by gram-negative organisms remains a major therapeutic challenge; mortality from this disease ranges from between 12 and 75% (8, 15, 16). Aminoglycosides administered parenterally are often used to treat this disease (11,14), although these antibiotics may penetrate the blood-brain barrier poorly. A favorable outcome in patients with meningitis infected with members of the family Enterobacteriaceae treated with aminoglycosides has been related to concomitant therapy with another effective antimicrobial agent, usually a beta-lactam (5, 16).Data on the penetration of gentamicin into cerebrospinal fluid (CSF) in humans following parenteral administration are fragmentary. Gentamicin concentrations in lumber CSF were measured by an radioenzymatic assay on four occasions in three neonates with gram-negative bacillary meningitis following a 2.5-mg/kg dose of gentamicin administered as an infusion over 30 min (5). Gentamicin was detectable (>1.0 ,xg/ml) in one of four CSF specimens. By a similar assay technique, no gentamicin could be detected in ventricular CSF of four hydrocephalic patients with ventriculitis following an intravenous infusion of undefined duration (20). Moellering and Fischer (17) measured gentamicin concentrations in lumbar CSF by using a microbiological assay in a 17-month-old child with meningitis. CSF concentrations of 0.5 and 0.3 ,ug/ml were observed 4 and 5 h, respectively, following an intramuscular dose of 2 mg/kg.In contrast to these results that indicated that aminoglycoside penetrated poorly, Eichenwald (7) tion of 1.6 ,ug/ml (range, 0.3 to 3.7 ,ug/ml) in lumbar CSF in 43 infants with gram-negative enteric meningitis, foll...