An in vitro pharmacodynamic model was used to simulate the in vivo pharmacokinetics of clarithromycin and 14-hydroxyclarithromycin in order to generate time-kill curves for three clinical isolates of Haemophilus influenzae (isolates 2019, 91-183, and 1746). Representative concentrations in serum or lung tissue and the pharmacokinetic parameters of clarithromycin and the 14-hydroxy metabolite, separately and in combination, were simulated for the time-kill studies. Amoxicillin-clavulanic acid was used as a control drug. The simulation of typical concentrations of the macrolides in serum in time-kill studies resulted in magnitudes of bacterial killing that were less than (for strains 2019 and 91-183, MICs = 4 mg/liter for clarithromycin and 14-hydroxyclarithromycin) or equal to (for strain 1746, MIC = 1 mg/liter for clarithromycin and 14-hydroxyclarithromycin) those observed in amoxicillin- clavulanic acid studies. When typical concentrations in lung tissue were simulated, total log decreases in bacterial counts were greater than those achieved with typical concentrations in serum and, in the case of strain 1746, exceeded the magnitude observed with the control drug. In each case, the time to 3-log-unit killing was longer for the macrolides than for amoxicillin-clavulanic acid. Time-kill curve analyses demonstrated the presence of synergy (defined as a 2-log-unit decrease in the CFU per milliliter between the combination and the most active constituent at any time point) for the combination of clarithromycin and 14- hydroxyclarithromycin at simulated concentrations in serum for one strain of H. influenzae (isolate 91-183). Synergism is likely bacterial strain specific, and the presence of synergy may be dependent on the antibiotic concentrations that are tested. Evaluation of the kill curve kinetics in terms of bactericidal rate for the various starting concentrations of clarithromycin did not result in a clear demonstration of either concentration-dependent or concentration-independent bactericidal activity.