Objectives Ferumoxytol is an alternative to gadolinium-based compounds as a vascular contrast agent for magnetic resonance angiography (MRA), particularly for patients with chronic kidney disease (CKD). However, dose-related efficacy data are lacking. We aimed to determine the optimal (minimum effective) dose of ferumoxytol for MRA in patients with CKD. Methods Ferumoxytol-enhanced MRA (FeMRA) was performed at 3.0 T in patients with CKD after dose increments up to a total of 4 mg/kg. Image quality was assessed by contrast-to-noise ratio (CNR) and signal-to-noise ratio (SNR) in the abdominal aorta and inferior vena cava. Quadratic regression analyses were performed to estimate the effects of dose increments on CNR and SNR. Results Twenty-three patients underwent FeMRA (mean age 60 SD 13 years, 87% men, 48% had diabetic nephropathy) with cumulative doses of 0, 1, 2, 3 and 4 mg/kg of ferumoxytol. On regression analyses, a parabolic relationship was observed between ferumoxytol dose and signal with progressive signal loss using doses exceeding 4 mg/kg. A dose of 3 mg/kg achieved ≥ 75% of predicted peak CNR and SNR and images were deemed of excellent diagnostic quality. Conclusions In patients with CKD undergoing FeMRA, a dose of 3 mg/kg provides excellent arterial and venous enhancement. The benefits of increasing the dose to a theoretically optimal value of 4 mg/kg appear to be negligible and likely of minimal, if any, diagnostic value. Key Points • Ferumoxytol is used off-label as an MRI contrast agent but dose-related data are lacking. • In patients with CKD requiring MR angiography, a dose of 3 mg/kg provides excellent vascular enhancement.