This study was undertaken to determine the temporal relationship between implementation of different interventions in an intensive care unit (ICU) and control of endemic nosocomial acquisition of extended-spectrum β-lactamase Enterobacteriaceae (ESBLE). This was a prospective observational study with time-series analysis of the monthly incidence of ESBLE and its predictors. In November 2007, after a 14-month baseline period, an intervention consisting of restriction of third-generation cephalosporins (3 GC) and increased use of alcohol-based hand rubs was implemented. In January 2008, an increased health care worker (HCW):patient ratio was also implemented. In March 2010, the ICU was closed, and patients were moved to a clean ICU. The first intervention resulted in global reduction in 3 GC and increased use of alcohol-based hand rub. A significant change in ESBLE incidence was observed in a full segmented univariate regression analysis (mean change in level, -0.91 ± 0.19; P < .0001). After ICU closure, there was a dramatic reduction in ESBLE acquisition. According to the multivariate model, the ICU closure was the main protective factor. Before ICU closure, an increase in the HCW:patient ratio of 0.1 point tended to be associated with a decreased risk of ESBLE acquisition (relative risk, 0.28; 95% confidence interval, 0.06-1.25; P = .09). This study shows that ICU closure was associated with, but not necessarily the reason for, control of ESBLE cross-transmission in a nonoutbreak setting. Environmental ESBE sources may play a role in cross-transmission.< Réduire