[en] OBJECTIVES: : To test the usefulness of procalcitonin serum level for the reduction of antibiotic consumption in intensive care unit patients. DESIGN: : Single-center, prospective, randomized controlled study. SETTING: : Five intensive care units from a tertiary teaching hospital. PATIENTS: : All consecutive adult patients hospitalized for > 48 hrs in the intensive care unit during a 9-month period. INTERVENTIONS: : Procalcitonin serum level was obtained for all consecutive patients suspected of developing infection either on admission or during intensive care unit stay. The use of antibiotics was more or less strongly discouraged or recommended according to the Muller classification. Patients were randomized into two groups: one using the procalcitonin results (procalcitonin group) and one being blinded to the procalcitonin results (control group). The primary end point was the reduction of antibiotic use expressed as a proportion of treatment days and of daily defined dose per 100 intensive care unit days using a procalcitonin-guided approach. Secondary end points included: a posteriori assessment of the accuracy of the infectious diagnosis when using procalcitonin in the intensive care unit and of the diagnostic concordance between the intensive care unit physician and the infectious-disease specialist. MEASUREMENTS AND MAIN RESULTS: : There were 258 patients in the procalcitonin group and 251 patients in the control group. A significantly higher amount of withheld treatment was observed in the procalcitonin group of patients classified by the intensive care unit clinicians as having possible infection. This, however, did not result in a reduction of antibiotic consumption. The treatment days represented 62.6 +/- 34.4% and 57.7 +/- 34.4% of the intensive care unit stays in the procalcitonin and control groups, respectively (p = .11). According to the infectious-disease specialist, 33.8% of the cases in which no infection was confirmed, had a procalcitonin value >1microg/L and 14.9% of the cases with confirmed infection had procalcitonin levels <0.25 microg/L. The ability of procalcitonin to differentiate between certain or probable infection and possible or no infection, upon initiation of antibiotic treatment was low, as confirmed by the receiving operating curve analysis (area under the curve = 0.69). Finally, procalcitonin did not help improve concordance between the diagnostic confidence of the infectious-disease specialist and the ICU physician. CONCLUSIONS: : Procalcitonin measuring for the initiation of antimicrobials did not appear to be helpful in a strategy aiming at decreasing the antibiotic consumption in intensive care unit patients.