Computed tomography angiography (CTA) is currently the gold standard imaging modality for anatomically characterizing canine hepatic vascular anomalies; with conventional, gadolinium‐enhanced MR angiography being less frequently utilized. However, both imaging modalities are limited by a brief, first pass peak of contrast medium in the vasculature that necessitates precisely timed image acquisition. A long‐acting purely intravascular magnetic resonance imaging (MRI) contrast agent, ferumoxytol, offers the potential to reduce complexity of magnetic resonance angiography (MRA) protocol planning by ensuring diagnostic contrast medium concentration in all the vessels that are targeted for imaging. Aims of this prospective, pilot, methods comparison study were to develop an optimized MRA protocol using ferumoxytol in dogs with hepatic vascular anomalies, perform a dose escalation trial to compare image quality with four‐dose regimens of ferumoxytol, and compare accuracy of vascular anatomic depiction based on the gold standard of CTA. Six dogs (10.7–36.1 kg) with portosystemic shunts (four intrahepatic left divisional shunts and two intrahepatic right divisional shunts) were recruited for inclusion in the study. A dose‐escalation trial was performed to compare image quality at four incremental dose levels of ferumoxytol (1, 2, 3, and 4 mg/kg) and to compare the accuracy of vascular anatomic detection to CTA. Ferumoxytol contrast‐enhanced MRA (CE‐MRA) at 4 mg/kg provided similar conspicuity of normal and abnormal vasculature compared to CTA with a minimal decrease in spatial resolution. Findings indicated that ferumoxytol holds promise for comprehensive, single breath hold CE‐MRA of all abdominal vessels in dogs with portosystemic shunts. Background information provided in this study can be used to support development of other future applications such as intracranial and cardiac MRA, real‐time imaging, flow quantification, and potentially sedated MRI imaging.