Inappropriate antibiotic use for pulmonary infiltrates is common in the intensive care unit (ICU). We sought to devise an approach that would minimize unnecessary antibiotic use, recognizing that a gold standard for the diagnosis of nosocomial pneumonia does not exist. In a randomized trial, clinical pulmonary infection score (CPIS) (Pugin, J., R. Auckenthaler, N. Mili, J. P. Janssens, R. D. Lew, and P. M. Suter. Diagnosis of ventilator-associated pneumonia by bacteriologic analysis of bronchoscopic and nonbronchoscopic "blind" bronchoalveolar lavage fluid. Am. Rev. Respir. Dis. 1991;143: 1121-1129) was used as operational criteria for decision-making regarding antibiotic therapy. Patients with CPIS </= 6 (implying low likelihood of pneumonia) were randomized to receive either standard therapy (choice and duration of antibiotics at the discretion of physicians) or ciprofloxacin monotherapy with reevaluation at 3 d; ciprofloxacin was discontinued if CPIS remained </= 6 at 3 d. Antibiotics were continued beyond 3 d in 90% (38 of 42) of the patients in the standard as therapy compared with 28% (11 of 39) in the experimental therapy group (p = 0.0001). In patients in whom CPIS remained </= 6 at the 3 d evaluation point, antibiotics were still continued in 96% (24 of 25) in the standard therapy group but in 0% (0 of 25) of the patients in the experimental therapy group (p = 0.0001). Mortality and length of ICU stay did not differ despite a shorter duration (p = 0.0001) and lower cost (p = 0.003) of antimicrobial therapy in the experimental as compared with the standard therapy arm. Antimicrobial resistance, or superinfections, or both, developed in 15% (5 of 37) of the patients in the experimental versus 35% (14 of 37) of the patients in the standard therapy group (p = 0.017). Thus, overtreatment with antibiotics is widely prevalent, but unnecessary in most patients with pulmonary infiltrates in the ICU. The operational criteria used, regardless of the precise definition of pneumonia, accurately identified patients with pulmonary infiltrates for whom monotherapy with a short course of antibiotics was appropriate. Such an approach led to significantly lower antimicrobial therapy costs, antimicrobial resistance, and superinfections without adversely affecting the length of stay or mortality.