Joint infections cause significant morbidity and mortality. Rapid diagnosis enables prompt initiation of appropriate antimicrobial therapy and surgical treatment. We conducted a systematic review and meta-analysis to evaluate the accuracy of genus- or species-specific polymerase chain reaction (PCR) in diagnosing joint infections. The literature databases were searched for articles from January 2010 to December 2022. The meta-analysis using the split component synthesis (SCS) method, included 20 studies with 2,457 adult participants. The pooled sensitivity, specificity, diagnostic odds ratio, and AUC of PCR were 49 % (95 % CI [37.9-60.2]), 95.7 % (95 % CI [91.6-97.8]), 21.32, and 0.82 respectively. Sensitivity was highest for sonicate fluid and lowest for periprosthetic tissue. The mean turnaround time to results was 4.7 hours (SD 1.1). PCR is a favourable option for diagnosing joint infections due to its rapid results, but it has low sensitivity. To enhance diagnostic yield, the test should be used in conjunction with other methods.