Magnetic resonance angiography (MRA) of renal vessels correlates well with conventional angiography (CA) and enables non-invasive assessment of renal vessels without nephrotoxic contrast. We aimed to identify the referral source and nature of the patient group undergoing renal MRA, and the impact of this test on their management. All renal MRA scans performed at the Royal Adelaide Hospital from 1 November 1997 to 31 December 2000 were reviewed (n = 121). Clinical data were obtained by case-note review or from treating physicians. MRA scans were with gadolinium enhancement, using a Siemens Vision 1.5 Tesla machine (Siemens, New York), with Visual Basic 33 software and 3D reconstruction. Nephrologists ordered the majority of renal MRA studies (64.5%). Indications for MRA included: (i). hypertension (91.3%), (ii). abnormal renal function (78.3%), (iii). other imaging suggesting renovascular disease (64.3%) and (iv) renal impairment with angiotensin-converting enzyme inhibition (18.3%). Eighty-seven MRA studies revealed renovascular abnormalities ('positive'). Over 50% of patients had three or more risk factors associated with vascular disease, with MRA positive in 76%. Localized renal artery stenosis was identified in 65 cases. In 40 of these, CA and further intervention was not undertaken, mainly due to presence of features of irreversible renal damage, low-grade stenosis or stable clinical parameters. CA was performed in 25 patients, all of whom had moderate- to high-grade (>60%)-stenosis on MRA. Revascularization was attempted in 21 of the 25 patients, with technical success in 17. In 22 patients, MRA identified non-localized abnormalities, most commonly diffuse arterial disease. None went on to angiography. Selective use of renal MRA in high-risk patients (identified by vascular risk factors or with standard renal imaging) can assist in avoidance of invasive, potentially nephrotoxic conventional angiography in up to 80% of cases. Those with lesions warranting treatment can then be selected for further intervention.