An in vitro pharmacokinetic model mimicking human serum drug concentrations, based on a dialyzer unit, was developed to study the efficacies of continuous infusion and intermittent administration of ceftazidime over a period of 36 h. The daily dose of ceftazidime was 300 mg/liter/24 h given either as a continuous infusion or as three bolus doses. The intermittent dosing regimen yielded peak and trough concentrations after the fourth dose of 92.3 (standard deviation, 8.0) and 1.4 (standard deviation, 0.9) mg/liter, respectively. Continuous administration yielded concentrations of approximately 20 mg/liter. To study efficacy, three Pseudomonas aeruginosa strains, ATCC 27853, CF4, and CF16, were used. The MICs of ceftazidime for these strains were 1, 4, and 16 mg/liter, respectively. Strain CF16 was killed initially during both regimens and then started to regrow. At the end of the fourth dosing interval, i.e., after 32 h, viable counts showed no difference between the regimens. Strains ATCC 27853 and CF4 were killed initially during both dosing schedules, and after the first dosing interval viable counts were similar. However, after the fourth interval, there was a marked difference between bacterial counts during continuous and intermittent infusion, being 2.2 and 2.8 log10, respectively, demonstrating a greater efficacy during continuous infusion. The results indicate that, in the absence of other factors, a sustained level of ceftazidime around or slightly above the MIC is not high enough to maintain efficacy over more than one (8-h) dosing interval. When sustained concentrations higher than four times the MIC are employed, continuous administration in this model is more efficacious than intermittent dosing.