Emerging evidence suggests that current fluoroquinolone dosing strategies may be inadequate to treat bloodstream infections caused by organisms classified as sensitive. This study sought to determine if differences in MICs for levofloxacin-susceptible gram-negative organisms correlate with differences in patient outcomes. A retrospective cohort study evaluated patients treated with levofloxacin for bloodstream infections caused by susceptible gram-negative organisms. Patients infected with gram-negative organisms for which MICs indicated susceptibility were categorized into three groups: those with organisms for which MICs were low (<or=0.25 mg/liter), intermediate (0.5 mg/liter), and high (1 or 2 mg/liter). Patients were evaluated for baseline similarity, all-cause mortality, and measurements of morbidity. A total of 404 patients with bloodstream infections caused by gram-negative organisms were identified. Of these patients, 312 were treated with levofloxacin and included in the analysis. No significant difference in all-cause mortality among the three groups was observed. The high-MIC group had a significantly longer average hospital stay postculture than the low- and intermediate-MIC groups (16.4 days versus 7.3 and 7.9 days; P < 0.01) and a significantly longer duration of infection (2.1 days versus 1.0 and 1.2 days; P < 0.001). Multivariate analysis adjusting for covariates revealed that a high MIC was associated with an increase of 5.67 days (95% confidence interval, 0.77 to 10.62 days; P = 0.02) in the mean length of stay postculture compared to the mean length of stay for the low-MIC group. Patients treated with levofloxacin for bloodstream infections caused by gram-negative organisms for which MICs were elevated, yet still in the susceptible category, had worse outcomes than similar patients infected with organisms for which MICs were lower. In vitro susceptibility classifications of fluoroquinolones for the treatment of bloodstream infections caused by gram-negative organisms require further study.