Diagnosis of chronic prosthetic joint infection (PJI) is often challenging. Painful prosthesis is frequently due to an infection but to diagnose it is somethimes difficult. All recent guidelines stress the central role of joint punction in diagnosis of PJI if the infection is not demonstrated. However which test on synovial fluid must be carried out is not so clearly defined. Total white blood cell count and differential leukocite count are usually considered useful in diagnosis but cut offs reported by different studies are quite different. Moreover this test needs a relatively large amount of fluid and blood contamination of it largely affects the result. What's more the synovial fluid WBC count may be unreliable in the setting of a metal-on-metal bearing or corrosion reaction. Routine cultures should be maintained between 5 and 14 days, their sensitivity appears low in chronic infection even if witholding antimicrobial therapy before the collection of the fluid can increase the likelihood of recovery an organism. Synovial leukocyte esterase can be performed as a rapid office or intraoperative point of care test using urinalysis strips. It is cheap and easy to perform, but the presence of blood in the sample can affect the result and it needs centrifugation. Recently a new test has been proposed to detect alfa-defensine in synovial fluid. It shows a high sensitivity and an exellent specificity. We performed 25 joint punctions on 25 patients with suspected PJI (enrollment is going on). Synovial fluid collected was tested for: leukocite esterase, WBC count and differential, colture in blood colture bottle for anerobe and aerobes (BacT/ALERT Biomerieux, inc) and detection of alfa-defensine level (Synovasure – Zimmer) In patients who underwent surgery at least 5 samples of periprotesic tissue were collected for microbiologic analysis and the removed implant was sonicated according with the methodic. Furthermore samples for frozen section were sent and a histologic examination was made according to the Moriewitz – Kerr classification. The MSIS criteria was utilized to classify the case as infected or not.