Aims. The aim of this study was to evaluate the diagnostic accuracy of the synovial alpha-defensin enzyme-linked immunosorbent assay (ELISA) for the diagnosis of prosthetic joint infection (PJI) in the work-up prior to revision of total hip (THA) and knee arthroplasty (TKA). Patients and Methods. Inclusion criteria for this prospective cohort study were acute or chronic symptoms of the index joint without specific exclusion criteria. Synovial fluid aspirates of 202 patients were analyzed and semiquantitative laboratory alpha-defensin ELISA was performed. Final diagnosis of PJI was established by examination of samples obtained during revision surgery. Results. Sensitivity and specificity of the alpha-defensin ELISA for PJI were 78.2% (95% confidence interval (CI) 66.7 to 88.5) and 96.6% (95% CI 93.0 to 99.3). Positive and negative predictive values were 89.6% (95% CI 80.6 to 97.8) and 92.2% (95% CI 87.5 to 96.1). The test remained false-negative in 22% of septic revisions, most of which were due to coagulase-negative staphylococci all occurring in either late-chronic or early-postoperative PJI. Conclusion. The routine use of synovial fluid alpha-defensin laboratory ELISA in the preoperative evaluation of symptomatic THAs and TKAs is insufficient to accurately diagnose PJI. Particularly in cases involving low-virulence organisms, such as coagulase-negative staphylococci, there remains a need for tests with a higher sensitivity. Cite this article: Bone Joint J 2019;101-B:970–977