ABSTRACTAntibiotics have been shown to influence the risk of infection with specificClostridium difficilestrains as well as the risk ofC. difficileinfection (CDI). We performed a retrospective case-control study of patients infected with the epidemic BI/NAP1/027 strain in a U.S. hospital following recognition of increased CDI severity and culture of stools positive byC. difficiletoxin immunoassay. Between 2005 and 2007, 72% (103/143) of patients with first-episode CDIs were infected with the BI strain by restriction endonuclease analysis (REA) typing. Most patients received multiple antibiotics within 6 weeks of CDI onset (median of 3 antibiotic classes). By multivariate analysis, fluoroquinolone and macrolide exposure was more frequent among BI cases than among non-BI-infected controls (odds ratio [OR] for fluoroquinolones, 3.2; 95% confidence interval [CI], 1.3 to 7.5; (P< 0.001; OR for macrolides, 5.2; 95% CI, 1.1 to 24.0;P= 0.04)). In contrast, clindamycin use was less frequent among the BI cases than among the controls (OR, 0.1; 95% CI, 0.03 to 0.4;P= 0.001). High-level resistance to moxifloxacin and azithromycin was more frequent among BI strains (moxifloxacin, 49/102 [48%] BI versus 0/40 non-BI,P= 0.0001; azithromycin, 100/102 [98%] BI versus 22/40 [55%] non-BI,P= 0.0001). High-level resistance to clindamycin was more frequent among non-BI strains (22/40 [55%] non-BI versus 7/102 [7%] BI,P= 0.0001). Fluoroquinolone use, macrolide use, andC. difficileresistance to these antibiotic classes were associated with infection by the epidemic BI strain ofC. difficilein a U.S. hospital during a time when CDI rates were increasing nationally due to the highly fluoroquinolone-resistant BI/NAP1/027 strain.