The recognition of a cluster of antibiotic-associated nosocomial Clostridium difficile disease (NCDD) caused by serotype C in a surgical ward led to a hospital-wide NCDD surveillance and control program. The initial step included: (a) gas-liquid chromatography screening of inpatients' diarrheal stools; (b) enteric isolation precautions, cohorting and terminal room disinfection in wards with a cluster of two or more NCDD cases per month. During a 12-month period, the quarterly incidence of NCDD remained unchanged and six new clusters of serotype C, K, and H infections occurred, giving a global incidence of 1.5/1,000 admissions. C. difficile spores were recovered from 36.7% surfaces of case patient rooms versus 6.7% in control rooms. More intensive control measures were evaluated: (a) culture screening of inpatients' diarrheal stools; (b) early therapy, enteric isolation precautions, and daily meticulous room disinfection for each sporadic NCDD case. Surface disinfection reduced the contamination level four-fold (p = 0.04). In the following 12 months, no cluster occurred and the incidence of NCDD fell to 0.3/1,000 admission (protective efficacy 73%, 95% confidence interval: 46-87%). These observations suggest that early therapy, isolation precautions, and surface disinfection, focused on patients with sporadic NCDD detected by active surveillance, can prevent nosocomial transmission of C. difficile.