OBJECTIVE The aim of this study was to evaluate the usefulness of serum procalcitonin (PCT) to differentiate between septic and nonseptic acute arthritis. METHODS Thirty-three patients who presented to our outpatient clinics with the symptoms of acute arthritis of unknown origin were enrolled in this study. We determined the serum concentrations of PCT and C-reactive protein and performed a white blood cell count. The definitive diagnosis was determined by microbiologic examination of a joint aspirate. RESULTS Microbiologic cultures showed that 15 patients had septic arthritis. In these patients the PCT levels were significantly increased compared with those in the 18 patients with nonseptic arthritis (mean +/- SD, 1.18 +/- 1.66 ng/mL v 0.078 +/- 0.073 ng/mL; P = .001). On the basis of the normal range (<0.5 ng/mL), the test sensitivity in our series was 53.3% with a specificity of 100%. C-reactive protein levels in patients with septic arthritis were also significantly elevated compared with the nonseptic group. Because in either group no value was within the normal range (<0.5 mg/dL), the specificity was 0%. CONCLUSIONS In this study serum PCT levels of patients with septic arthritis were significantly higher than those of patients with nonseptic arthritis (P = .001). However, the sensitivity when referencing the normal upper limit of PCT (0.5 ng/mL) is not high enough to establish a diagnosis based exclusively on serum parameters. This reflects the results of other studies examining localized infections without generalized bacteremia. The data suggest that a specific PCT cutoff level for localized infections could be helpful in the future. METHODS Level III, diagnostic study of nonconsecutive patients without consistently applied reference gold standard.