To determine the potentials and limitations of a recently developed algorithm for virtual intravascular endoscopy (VIE) based on 3-dimensional (3-D) magnetic resonance (MR) data sets.The data from 6 patients derived from gadolinium-enhanced MR angiography of the renal arteries were reviewed. All patients had unilateral or bilateral renal arterial pathologies (stenoses n = 2, aneurysms n = 4). Imaging was performed on a 1.5-T scanner using a 3-D gradient-echo MR sequence. The imaging parameters were as follows: TR/TE/flip angle 3.9/1.9/40 degrees, matrix 256 x 192, slice thickness 1.5 to 2 mm, FOV 32 cm, and 48 slices. Image acquisition time was 28 seconds under apnoeia conditions. A total of 60 ml 0.5 molar Gd-DTPA was injected during the scan.3-D data sets could be obtained in all patients and postprocessed for VIE. The intraluminal vessel wall was depicted with high clarity in all cases. All pathologies that were not intraparenchymal could be easily seen. Limitations to the technique include the image quality of the original data set, use of the ideal threshold to minimise intraluminal artifacts, and a complicated prescription sequence.We have shown that VIE can be consistently performed in the renal arteries using MR data sets acquired with a contrast enhanced 3-D gradient-echo technique. It provides a hitherto unused approach to viewing 3-D vascular MR data sets.