The relative effectiveness of bolus vs. constant intravenous administration of equivalent doses of penicillin G in killing bacteria in vivo was studied in a rabbit model of meningitis due to Streptococcus pneumoniae. Samples of cerebrospinal fluid (CSF) and serum were obtained from 30 rabbits at intervals of~8 hr after treatment for determination of antibiotic concentrations and titers of viable bacteria in the CSF. When penicillin G was given by continuous infusion (105 units/hr after an initial lOS-unit loading dose), concentrations of drug in serum and CSF reached a steady state in 1 hr. With intermittent bolus administration of 4 x 105 units every 4 hr, higher peak and lower trough concentrations were achieved, and these concentrations paralleled those in the CSF. Although an initial acceleration in bactericidal rate was observed with the bolus infusion between the first and second hour of therapy, after the second hour the rate of bacterial killing was identical for the two methods of administration. The duration of therapy required for sterilization of the CSF was dependent only on the bacterial count before treatment and not on the mode of drug administration. The effect of single bolus intravenous administration of ampicillin was examined in experimental pneumococcal meningitis. Ampicillin was given at various dosages (3.25-62.5 mg/kg), and frequent samples of CSF were obtained for determination of concentrations of pneumococci and ampicillin. A long postantibiotic effect was observed in the CSF of all animals, and this effect consistently was longer than that observed in vitro.The question of the most effective mode of administration of (3-lactam antibiotics for treatment of severe bacterial infections is controversial. Commonly used regimens are largely empirical and are based on pharmacokinetic parameters (often only available from studies done with normal volunteers), knowledge of susceptibility characteristics of the suspected pathogen, and tradition. Parenteral therapy is preferred, but unpredictable absorption (e.g., shock) and clotting abnormalities may make the im route unreliable. Continuous iv administration produces a relatively fixed concentration of the antibiotic over time. Intermittent, or bolus, administration produces high initial con-