Amoxicillin subinhibitory concentrations produced 100% survival in passively immunized mice infected with non-amoxicillin-susceptible, poorly or highly encapsulated Streptococcus pneumoniae strains (2, 6), with negligible values of time that serum levels exceeded the MIC. We explored this phenomenon and its modulation by capsular production with levofloxacin against the same serotype 6B Streptococcus pneumoniae strain (6) (MIC of levofloxacin = 32 μg/ml) with two types of infecting inocula. (i) For the poorly encapsulated (PE) phenotype, the microorganism was grown in Todd-Hewitt broth supplemented with 0.5% yeast extract (Difco, Detroit, Mich.) until an absorbance of 0.3 at 580 nm (UV-visible spectrophotometer, Shimadzu UV-1203, Japan) was reached. (ii) For the highly encapsulated (HE) phenotype (4), after serial passages in mice, the microorganism was grown three times in Todd-Hewitt broth supplemented with 0.5% yeast extract (Difco, Detroit, Mich.) and enriched with 5% fetal bovine serum until an absorbance of 0.3 at 580 nm (UV-visible spectrophotometer, Shimadzu UV-1203, Japan) was reached. Eight-to 12-week old female BALB/c mice weighing 19 to 22 g were used. The challenge dose with the PE and the HE inocula (2, 6) was 4 × 108 CFU/ml. Previously described methods (6) were followed for hyperimmune serum production and determination of protection with and without levofloxacin doses decreasing on a twofold basis from 25 mg/kg of body weight. Groups of 10 animals per dose were used. Experiments were carried out in duplicate. Treatment was initiated 1 h after the intraperitoneal challenge, and a second dose was administered 24 h later. Levofloxacin concentrations were determined by bioassay using Escherichia coli ATCC 25922 in pooled sera from five animals per sampling time (predosing, 15 min, 30 min, 1 h, 2 h, and 4 h). Drug concentrations were analyzed by a noncompartmental approach using the WinNonlin Professional program (Pharsight, Mountain View, Calif.). Survival curves were obtained by the Kaplan-Meier method. An ordinal log-rank test was used to compare different study groups. Due to multiple comparisons, a P value of ≤0.001 was considered significant. Concentrations (μg/ml) of levofloxacin in serum obtained after a single 25-mg/kg dose were 144.54 at 15 min, 120.22 at 30 min, 4.67 at 1 h, 0.23 at 2 h, and undetectable at 4 h. Maximum concentration and area under the curve (AUC) were 144.54 μg/ml and 84.84 μg · h/ml. Table Table11 shows survival rates. No differences (P = 0.85) between the PE and the HE models were found with nonimmune serum or placebo controls. In the PE model, differences in survival rates between immunized and nonimmunized animals were nonsignificant (P = 0.03) with 6.25 mg/kg levofloxacin but significant (P < 0.0001) with the 12.5-mg/kg dose. Significant differences (P < 0.0001) were found, with higher survival rates in the PE than the HE model (0% from day 2 onwards), for each treatment regimen. TABLE 1. Survival rates produced by three levofloxacin doses over a 7-day follow-up period with both types of infecting inocula (PE and HE) in normal mice and previously immunized mice An AUC/MIC ratio of 25 to 30 has been classically related to favorable outcomes in humans infected with S. pneumoniae (1) despite data supporting lower values needed (5). Lower values are needed in rodents (3). In the present study, ratios of maximum concentration to MIC and AUC to MIC of 4.5 and 2.7, respectively, produced efficacy (80% survival) in the PE model. These values were not enough to produce efficacy when the strain was highly encapsulated (HE model), where an increase in capsule-associated virulence was noted. Human natural infections by S. pneumoniae occur with highly capsulated strains, suggesting that much higher AUC/MIC ratios are needed in natural infections. We express our gratitude to L. Alou (IPM, Madrid, Spain) for the statistical analysis.