BACKGROUND: Bloodstream infection (BSI) is associated with a high mortality rate. Since the origin of infection is demonstrated in approximately 2/3rds of cases, early and established biomarkers are warranted. We evaluated the clinical performances of automated procalcitonin (PCT) and C-reactive protein (CRP) assays for the quantitative detection of BSI. Analytical performance of the VIDAS(R) B.R.A.H.M.S PCT assay (bioMerieux, France) was assessed and also compared with the semi-quantitative PCT-Q test (B.R.A.H.M.S Aktiengesellschaft, Germany). METHODS: We prospectively included consecutive patients divided into 3 groups at the Dong-A University Medical Center. Patients were categorized according to the criteria of the American College of Chest Physicians/Society of Critical Care Medicine Consensus Conference (ACCP/SCCM), and also on the basis of catheter-associated bacteremia. RESULTS: A total 77 patients were enrolled. All mean values of PCT and PCT-Q were consistent with the reference value. Measured PCT concentrations showed good linearity (r=0.983). The between-run, within-run, and total imprecisions were below 5%. The PCT levels in gram-negative bacteremia were significantly higher than those in gram-positive bacteremia. Furthermore, the PCT concentrations were significantly different among non-infection, bacteremia, sepsis, severe sepsis, and septic shock groups. Our study showed that PCT >0.3 ng/mL had 95.0% sensitivity and 97.3% specificity, whereas CRP >5.46 mg/dL had 85.0% sensitivity and 86.5% specificity for diagnosing sepsis. CONCLUSIONS: We suggest that, compared with CRP, PCT is a better diagnostic and discriminative biomarker of sepsis categorized according to the ACCP/SCCM. Moreover, catheter-associated bacteremia could be discriminated from sepsis using PCT concentration.