ObjectiveTo determine the effect of restriction of third-generation cephalosporin use on antibiotic resistance and the outcome of patients with infection.DesignA prospective, before–after comparative study.SettingA general intensive care unit with 14 beds at a university-affiliated teaching hospital.PatientsAll patients admitted to the intensive care unit within 2 yrs.InterventionsA new antibiotic treatment strategy was implemented during phase II. All patients with confirmed or suspected Gram-negative bacterial infections were treated mainly with antibiotics other than third-generation cephalosporins.Measurements and Main ResultsAntibiotic resistance among common Gram-negative bacilli and infection-related hospital mortality during phase I were compared with phase II. A 26.6% reduction in third-generation cephalosporin use (from 168.2 ± 48.0 to 123.5 ± 39.3 g/month,p= .021), accompanied by a 277.7% increase in cefepime use (from 10.3 ± 19.2 to 38.9 ± 31.7 g/month,p= .014) occurred in phase II compared with phase I. This was accompanied by a significant decrease in reduced susceptibility of Gram-negative bacilli to third-generation cephalosporins (p< .05), mainly because of the improved susceptibility ofEscherichia coliandKlebsiellaspecies (p< .05). Infection-related hospital mortality was significantly lower in phase II (19.3% vs. 36.3%,p= .014). Multiple logistic regression analysis demonstrated lower respiratory tract infection, the status of immunocompromise, and continuous veno-venous hemofiltration as independent risk factors for infection-related hospital mortality (p< .05), whereas infection withE. coliorKlebsiellaspecies (p= .039) and restriction of third-generation cephalosporin use (p= .025) were associated with a significantly lower mortality rate.ConclusionsRestriction of third-generation cephalosporin use may improve the antibiotic susceptibility and reduce infection-related hospital mortality in critically ill patients.