The ideal approach to diagnosis and management of patients with ventilator-associated pneumonia (VAP) has not been defined. Trauma patients with suspected VAP underwent bronchoscopy for collection of quantitative bacterial cultures and were randomized to receive either prompt empiric broad-spectrum therapy modified by culture results (intervention) or antibiotic therapy based only on results of positive cultures (control). Among 98 evaluable patients, 75 were victims of motor vehicle accidents or gunshot wounds. Seventy-two had positive bronchoscopic cultures and 18 had positive confirmatory blood or pleural fluid cultures. Among intervention patients with VAP (n = 44), the median number of days of mechanical ventilation was 17, versus 23 in control patients with VAP (n = 28; p = 0.11). Intervention patients required a median of 22 days of intensive care, while control patients required 31 (p = 0.07). Intervention patients were hospitalized a median of 37 days, control patients for 53 days (p = 0.12). Hospital charges were $106,071 in patients with VAP who received early empiric antibiotics, versus $162,413 in control patients (p = 0.06). Mortality did not differ between the two groups (18% vs. 23% overall; 21% vs. 25% in VAP patients only). Patients with VAP had a median C-reactive protein (CRP) of 24.7 mg/dL versus 18.4 mg/dL in those without VAP (p = 0.003). The receiver-operator characteristics curve area for CRP was higher than that for blood leukocyte count in predicting VAP (0.68 ± 0.06 vs. 0.45 ± 0.06; p < 0.003). Among VAP patients, serial measurements during antibiotic therapy showed a reduction of CRP from baseline to levels similar to those of patients without VAP (posttreatment level, 9.8 mg/dL; p < 0.0001). Day 14 CRP levels were lower in survivors (6.4 mg/dL vs. 14.3 mg/dL; p = 0.025). In trauma patients with suspected VAP, specific antimicrobial therapy refined by results of quantitative bronchoscopic cultures and prompt empiric antimicrobial therapy are associated with similar outcomes. CRP measurement in this setting is a potentially worthwhile marker of the presence of VAP and its response to therapy.