To increase the sensitivity of 3-dimensional fluid-attenuated inversion recovery (3D-FLAIR) to low concentration gadolinium (Gd)-based contrast medium, we optimized sequence parameters on a phantom and evaluated the optimized sequence in patients suspicious for endolymphatic hydrops. All scans were performed on a 3-tesla magnetic resonance (MR) unit using a 32-channel head coil. We optimized sequence parameters using a phantom filled with diluted Gd and compared the optimized protocol with 3D-FLAIR using conventional turbo spin echo sequence (3D-FLAIR-CONV). Nine patients underwent scanning using the newly optimized sequence and 3D-FLAIR-CONV 4 hours after double-dose administration of intravenous Gd. We subjectively scored separation of endo- and perilymph space and measured contrast-to-noise ratio (CNR) between endo- and perilymph. The optimized sequence in the phantom study consisted of: repetition time, 9000 ms; echo time, 540 ms; inversion time, 2400 ms; low constant readout flip angle, 120 degrees in the later part of the echo train. Image contrast became heavily T(2)-weighted (hT(2)W-3D-FLAIR). In patients, we recognized endolymphatic space for both the cochlea and vestibule significantly better by hT(2)W-3D-FLAIR than 3D-FLAIR-CONV (P<0.01). The mean CNR of the new method was also better than that of 3D-FLAIR-CONV (P<0.01). The newly optimized hT(2)W-3D-FLAIR was more sensitive than the previous method to low concentration of Gd. Visualization of the endolymphatic space by double-dose administration of intravenous Gd would be more reliable using hT(2)W-3D-FLAIR.