BACKGROUND This paper reports the use of different vascular access devices and the incidence of intravascular catheter-related infection (CRI) in patients receiving intravenous antibiotics for infective endocarditis (IE). AIM To examine whether rates of infection vary with type of vascular access device, and assess the impact of CRI on mortality in IE. METHODS A prospective observational service evaluation of all inpatients who received intravenous antibiotics for IE was performed. In total, 114 inpatients were evaluated. All cases of CRI [including exit-site infection, intravascular catheter-related bloodstream infection (CRBSI) and mortality] were recorded. Tunnelled and non-tunnelled central venous catheters (CVCs), and peripherally inserted cannulae were used for antibiotic delivery. FINDINGS There were 15 episodes of CRI, 11 of which were CRBSI (all associated with CVC use). The remainder comprised uncomplicated exit-site infections. Use of tunnelled CVCs [hazard ratio (HR) 16.95, 95% confidence interval (CI) 2.13-134.93; P = 0.007] and non-tunnelled CVCs (HR 24.54, 95% CI 2.83-212.55; P = 0.004) was associated with a significantly increased risk of CRI. Risk of mortality increased significantly with Staphylococcus aureus as the cause of IE (P < 0.001) and CRBSI (P = 0.034). CONCLUSION Risk of CRI in patients with IE is linked to the type of vascular access device used. Rates of CRBSI were greatest with CVCs, while peripheral venous cannulae were not associated with CRBSI or serious sequelae. Many patients (40%) tolerated complete treatment courses delivered via peripheral cannulae. These findings confirm the importance of device selection in reducing the risk of CRI; a potentially modifiable variable that impacts on outcome and mortality in IE.