Background Staphylococcus aureus bacteremia in patients with cardiac implantable electronic devices (CIEDs) is often associated with infective endocarditis (CIED-IE). The CIED-IE diagnosis is syndromic. Diagnostic uncertainty is common. Frequently, these patients are classified as having possible CIED-IE, resulting in guideline-noncompliant treatment and heterogeneous outcomes. Improved outcomes require accurate diagnoses. In these patients, we evaluated whether metagenomic sequencing of microbial cell-free DNA (mcfDNA) in serial plasma specimens could improve diagnostic precision.Methods We studied 16 patients with staphylococcal bacteremia who were classified with definite or possible CIED-IE and recommended for device removal, if there was a positive blood culture within 7 days and no concurrent deep infection. Plasma specimens obtained at consent, before extraction, and during 96 hours after extraction underwent metagenomic sequencing and quantification of staphylococcal mcfDNA.Results In 10 of 11 patients with definite CIED-IE, mcfDNA persisted during antibiotic therapy for prolonged durations (median, 11 days [interquartile range {IQR}, 7.5-15 days]). In these cases, mcfDNA concentration in plasma obtained early after lead extraction increased significantly and thereafter decreased rapidly. In 5 cases of possible CIED-IE, mcfDNA was undetectable after 6 days (IQR, 5.5-7.5 days) of antibiotic therapy and remained undetectable after CIED extraction. These mcfDNA patterns differ significantly (P = .001), suggesting 2 distinct patient populations: 1 with definite CIED-IE and 1 without lead infection.Conclusions If confirmed, these mcfDNA patterns can serve as biomarkers, together with clinical features, to improve precision in diagnosing or rejecting S. aureus CIED-IE. Strategically timed mcfDNA testing before and after CIED extraction may aid in planning therapy. Among patients with cardiac implantable electronic devices experiencing staphylococcal bacteremia, quantitative microbial cell-free DNA detected in plasma obtained serially during antibiotic therapy, before and after device extraction, yielded unique biomarker patterns indicative of the presence or absence of device infection.