For some pneumococci the fluoroquinolone MICs are low but the mutant prevention concentrations (MPCs) are high; this difference defines in vitro the mutant selection window (MSW). We investigated in vivo the bacterial reduction and the occurrence of resistant mutants with moxifloxacin (MFX; 400 mg once daily) or levofloxacin (LVX; 500 mg twice daily) in treatments similar to those in humans with experimental pneumonia due to pneumococci (expPP) exhibiting various MICs and MPCs. The MIC/MPC for MFX and LVX and genotypes were as follows: strain 16089, 0.125/0.125 and 0.5/0.5 (wild type); strain MS1A, 0.25/0.25 and 1/2 (efflux); strain MS2A, 0.25/4 and 1.75/28 (parC79); strain MR3B4, 0.25/4 and 2/32 (parC79); strain M16, 0.5/2 and 8/32 (parC83); strain Gyr-1207, 1.5/3 and 8/16 (gyrA); and strain MQ3A, 4/4 and 16/64 (parC and gyrA). Both drugs were efficient with wild type-expPP, but only MFX was efficient with efflux-expPP. No bacterial reduction was observed for parC-expPPs due to mutants observed in 18 to 100% of animals, depending on the strain and the drug tested. These mutants showed unbound area under the concentration-time curve and MICs of from 50 to 164 for MFX. The in vivo pharmacodynamic boundaries of the MSW were different for MFX and LVX. We conclude that, after LVX or MFX treatment, mutants occur in vivo if there is a preexisting parC mutation, since the drug concentrations fall below the MPCs of these strains. Since the MPC determination cannot be routinely determined, these phenotypes or genotypes should be detected by simple tests to guide the therapeutic options.