OBJECTIVE When revision surgery is needed in total knee arthroplasty (TKA) the most frequent reasons are aseptic loosening (AL) and periprosthetic joint infection (PJI). However preoperative distinction between AL and PJI remains challenging. Aim of this study is to determine the incidence of PJI in patients with suspected AL after TKA and to evaluate a diagnostic algorithm for reliable differential diagnosis. METHODS In this study a total of 149 symptomatic patients with radiographic signs of prosthetic loosening and suspected AL were included. Preoperatively all patients underwent a standardized diagnostic algorithm. For each patient demographics, as well as the results of laboratory and microbiological testing were collected from the medical records. RESULTS Of the included patients 117 (78.5%) were diagnosed with AL and 32 (21.5%) with PJI. The latency period from primary arthroplasty to the presentation with symptomatic implant loosening was significantly shorter for PJI compared to AL (p < 0.05). The initial CRP values were significantly higher in patients with PJI compared to patients with AL (p < 0.05). Elevated count of white blood cells or percentage of neutrophils within the synovial fluid support the diagnosis of PJI. The sensitivity of synovial cell count (CC) count for PJI in patients with radiographic signs of loosening was 0.84 (CI 0.81-0.87) with a specificity of 0.96 (CI 0.92-0.98). The single best measure for the diagnosis of PJI was synovial fluid cultures with a specificity of 1, however this measure provides poor sensitivity. CONCLUSIONS Patients with radiographic signs of loosening in TKA need thorough diagnostics. Information about primary TKA, serological testing, and results of joint aspiration can rule out a PJI in most cases.