Hospital-acquired pneumonia (HAP) is associated with a high mortality rate and a substantial economic burden. Many clinical and economic studies do not differentiate between HAP in the intensive care unit (ICU) versus on the ward, even though they are associated with different pathogens, treatment approaches, mortality and costs. For nosocomial infections, the effectiveness of therapy is dependent on the pathogenic organism and susceptibility to prescribed antibiotics, factors that are specific to each individual institution. OBJECTIVES: To develop decision analytic models for current and anticipated antimicrobial regimens for the initial treatment of ICU and non-ICU HAP, that incorporate site-specific pathogen and susceptibility profiles, from the hospital perspective. METHODS: Two decision tree models were constructed, one for ICU and the other for non-ICU HAP. The probabilities for the decision analysis model were derived from a meta-analysis of randomized, controlled clinical trials and data from our hospital population. Antimicrobial susceptibilities were obtained from the literature and local data. The comparators included in the baseline analysis were cefotaxime (CFX), ceftazidime (CTZ), ceftriaxone (CTR), ciprofloxacin (CIP), imipenem (IMP), and cefazolin + gentamicin (C+G). Drug acquisition, pharmacy, nursing, and hospitalization costs were included in the analysis. Hospitalization costs were determined from local case-costing data. Outcomes were measured as success, failure and death. RESULTS: For ICU HAP, C+G dominated over IMP. The incremental cost-effectiveness ratios for CTZ and CIP were $66,087/success and $49,099/success, respectively. For non-ICU HAP, C+G dominated over CTR and CFX. The incremental ratio for CTZ was $142,500/success. Sensitivity analyses did not substantially alter the results. CONCLUSIONS: Antimicrobial susceptibility is a clinically important determinant of efficacy that should be included in economic analyses of HAP and other infections.