T2-weighted high-resolution gradient and fast spin echo sequences are widely used as an alternative or adjunct to contrast-enhanced T1-weighted temporal bone imaging. However, to date no systematic comparison has been presented. The purpose of this work is to identify optimal acquisition parameters and to compare volume gradient and fast spin echo techniques. Signal intensities and scan efficiency were computed for gradient echo segment-interleaved motion-compensated acquisition into steady state (SIMCAST), standard fast spin echo (FSE), and fast recovery fast spin echo (FR-FSE). Computations were compared with inner ear images acquired with cubic voxel sizes of 0.35-0.40 mm(3)in 5-8 minutes. Given otherwise identical conditions, the FR-FSE sequence produces images with improved SNR in shorter scan times than standard FSE. For FR-FSE, the scan efficiency is optimal for specific pairs of TR and echo train length, eg, 400 ms/8, 735 ms/16, and 2,050 ms/48. FR-FSE images with large TR and echo trains, while achieving better SNR, are severely compromised by blurring. Imaging with echo train lengths of 16-24 and TR of 800-1,200 ms is a good compromise, and FR-FSE signal-to-noise ratio (SNR) and scan efficiency become comparable to SIMCAST. In vivo image quality is excellent with both FR-FSE and SIMCAST, but SIMCAST images have slightly higher SNR and are significantly more crisp. J. Magn. Reson. Imaging 2000;12:814-825.