The in vivo efficacies of pefloxacin, a new fluoroquinolone, and amikacin-ceftazidime were compared in 50 rabbits with experimental aortic endocarditis caused by Pseudomonas aeruginosa. Animals were randomly chosen to receive 4 or 10 days of no therapy (controls), pefloxacin (40 mg/kg [body weight] per day, intramuscularly [i.m.]), or amikacin (30 or 80 mg/kg per day, i.m.)-ceftazidime (150 mg/kg per day, i.m.). Pefloxacin and both amikacin regimens significantly reduced vegetation bacterial densities compared with controls at days 4 and 10 of treatment (P less than 0.0005). By day 10 of therapy, between 33 and 40% of vegetations from amikacin-ceftazidime recipients contained ceftazidime-resistant bacteria (MICs, greater than 25 micrograms/ml); nitrocefin agar overlay confirmed that these ceftazidime-resistant variants were constitutive overproducers of beta-lactamase. At therapy days 4 and 10, approximately 30% of vegetations sampled from pefloxacin recipients contained bacteria for which pefloxacin MICs were four- to eightfold higher than the MIC for the parental strain used to initially induce endocarditis (MIC, 0.19 microgram/ml). These variants also exhibited increases in ciprofloxacin and ticarcillin MICs, as well as pleotropic resistance to chloramphenicol (but not to amikacin, ceftazidime, or tetracycline). Amikacin-ceftazidime, as well as pefloxacin, was effective in this model of aortic pseudomonal endocarditis. However, in vivo development of ceftazidime resistance and step-ups in pefloxacin MICs among intravegetation isolates were associated with inability to completely eradicate P. aeruginosa from aortic vegetations.