Clostridium difficile-associated colitis is an increasing cause of morbidity and mortality in hospitalized patients, with high relapse rates following conventional therapy. We sought to determine the efficacy of rifaximin, a novel nonabsorbed antibiotic, in the hamster model of C. difficile-associated diarrhea (CDAD). Hamsters received clindamycin subcutaneously and 24 h later were infected by gavage with one of two C. difficile strains: a reference strain (VPI 10463) and a current epidemic strain (BI17). Vancomycin (50 mg/kg of body weight) or rifaximin (100, 50, and 25 mg/kg) were then administered orally for 5 days beginning either on the same day as infection (prevention) or 24 h later (treatment). Therapeutic effects were assessed by weight gain, histology, and survival. We found that rifaximin was as effective as vancomycin in the prevention and treatment of colitis associated with the two C. difficile strains that we examined. There was no relapse after treatment with vancomycin or rifaximin in hamsters infected with the BI17 strain. Hamsters infected with the VPI 10463 strain and treated with rifaximin did not develop relapsing infection within a month of follow-up, whereas the majority of vancomycin-treated animals relapsed (0% versus 75%, respectively; P < 0.01). In conclusion, rifaximin was found to be an effective prophylactic and therapeutic agent for CDAD in hamsters and was not associated with disease recurrence. These findings, in conjunction with the pharmacokinetic and safety profiles of rifaximin, suggest that it is an attractive candidate for clinical use for CDAD.