The aim of this study was to assess the feasibility of magnetic resonance renography (MRR) using gadolinium diethylenetriaminepentaacetic acid (Gd-DTPA) in comparison with conventional radionuclide renography (RR) using technetium-99m-DTPA (99mTc-DTPA). MRR has many advantages over RR, including lack of ionising radiation, increased spatial resolution, and visible background anatomy. By optimising the pulse sequence, we developed an MRR protocol in which signal intensity is linear with Gd-DTPA concentration over a clinically relevant range. Twenty-nine patients and a volunteer were studied using this protocol. Magnetic resonance renography was performed using three different doses of Gd-DTPA: 0.1 mmol kg-1 (n = 13), 0.05 mmol kg-1 (n = 7), and 0.025 mmol kg-1 (n = 9). Each patient was also assessed using radionuclide renography. The resulting renograms were assessed in terms of time to peak signal intensity, signal decrease after peak, and kidney function ratios calculated from both the areas underneath and the slopes of the uptake curves. We have shown that the MR renograms obtained using low dose Gd-DTPA correlate best with the radionuclide renograms. Remaining discrepancies may be explained by variations in the injection procedures (hence in arterial input functions) and the limited coverage of the three MRR slices compared to the whole body projection of RR. Furthermore, at high local concentrations, signal becomes independent of T1 and is dominated by T2.