In this study, we determined the efficacy of various dosing regimens for erythromycin and azithromycin against four pneumococci with different susceptibilities to penicillin in an in vitro pharmacokinetic model and in a mouse peritonitis model. The MIC was 0.03 microg/ml, and the 50% effective doses (determined after one dose) of both drugs were comparable for the four pneumococcal strains and were in the range of 1.83 to 6.22 mg/kg. Dosing experiments with mice, using regimens for azithromycin of one to eight doses/6 h, showed the one-dose regimen to give the best result; of the pharmacodynamic parameters tested (the maximum drug concentration in serum [Cmax], the times that the drug concentration in serum remained above the MIC and above the concentration required for maximum killing, and the area under the concentration time curve), Cmax was the best predictor of outcome. The bacterial counts in mouse blood or peritoneal fluid during the first 24 h after challenge were not correlated to survival of the mice. The serum concentration profiles obtained with mice for the different dosing regimens were simulated in the in vitro pharmacokinetic model. Here as well, the one-dose regimen of azithromycin showed the best result. However, the killing curves in vivo in mouse blood and peritoneal fluid and in the vitro pharmacokinetic model were not similar. The in vitro killing curves showed a decrease of 2 log10 within 2 and 3 h for azithromycin and erythromycin, respectively whereas the in vivo killing curves showed a bacteriostatic effect for both drugs. It is concluded that the results in terms of predictive pharmacodynamic parameters are comparable for the in vitro and in vivo models and that high initial concentrations of azithromycin favor a good outcome.