Various three-dimensional fast spin echo (3D-FSE) sequences are used for non-contrast magnetic resonance angiography (MRA). Differences in the ability to detect vascular stenosis using these sequences, however, have not yet been evaluated. The purpose of this study is to evaluate the usefulness of each sequence for the detection of vascular stenosis by using a vascular phantom. The phantom consisting of silicon tubes with 30% and 70% stenosis of luminal diameter and fluids close to T₂ value of blood were used for the study. Non-contrast MRA with half-Fourier acquisition single-shot turbo spin echo (HASTE)-noncontrast magnetic resonance angiography of arteries and veins (NATIVE), sampling perfection with application optimized contrasts using different flip angle evolutions (SPACE)-NATIVE, fresh blood imaging (FBI) and triggered angiography non contrast enhanced (TRANCE) sequences was performed by using the phantom which can be varied in terms of the steady flow velocity. Each stenosis was quantitatively estimated by the stenosis index (SI) calculated from the signal intensities on acquired images. The signal intensity of the non-stenotic vascular site markedly decreased at more than a flow rate of 20 cm/s in all sequences. Significant decrease in the signal intensity was observed in the distal point from the stenosis area on these images acquired by using HASTE-NATIVE and FBI sequences. FBI and TRANCE sequences showed a more accurate SI for 30% stenosis than HASTE-NATIVE and SPACE-NATIVE sequences. SI for 70% stenosis was overestimated in all sequences at 5 cm/s of diastolic flow rate. In conclusion, the ability to detect vascular stenosis on non-contrast MRA image using 3D-FSE sequences depends on the image quality during diastolic phase in the cardiac cycle. FBI and TRANCE sequences are useful to detect the mild arterial stenosis.