Study objectives: To compare the plasma concentration of C-reactive protein (CRP) with traditional markers for diagnosis of bacterial pneumonia in patients with suspected aspiration. Design: Prospective, nonrandomized, controlled study of consecutive hospital admissions. Setting: Toxicology ICU in a university hospital. Patients or participants: Acutely poisoned comatose patients admitted to the hospital with suspicion of aspiration pneumonia. Interventions: Distal protected catheter sampling per fiberoptic bronchoscopy and bacteriologic culture were employed as a standard to detect the bacterial component of suspected aspiration pneumonia. Plasma CRP concentrations, temperature, and WBC count were measured on hospital day 1. Measurements and results: Sixty-six patients were evaluated. Thirty-two had bacterial contamination by positive culture (greater than or equal to 10(3) cfu/mL). Multiple receiver-operating characteristic (ROC) curves were used to compare each parameter for detection of infection secondary to aspiration. The ROC curve of CRP concentrations showed that a CRP > 75 mg/L is associated with bacterial contamination with a sensitivity of 87%, specificity of 76%, positive predictive value of 78%, and negative predictive value of 87%. ROC curves of temperature and WBC count demonstrated poor diagnostic value of these markers in indicating the bacterial component of suspected aspiration pneumonia. Conclusions: Early measurement of CRP is useful for the diagnosis of aerobic bacterial content of aspiration pneumonia and perhaps in determining the need for invasive bacteriologic sampling. Temperature and WBC count are poor indicators of bacterial infection of aspiration pneumonia in poisoned patients.