In order to improve head and neck radiotherapy treatment planning, a series of 3D fat-suppressed T2-weighted sequences were developed with varying pulse sequence parameters. One non-fat-suppressed and five fat-suppressed sequences were acquired on five patients, and on each image, four structures were segmented by six radiation oncology physicians (observers). Robust and comprehensive analysis was performed to assess qualitative and quantitative image quality metrics. This included geometric distortion, SNR and CNR measurements, structure conspicuity, interobserver segmentation variability, and qualitative image quality rankings. The results of these were used to determine the optimal fat-suppressed sequence for head and neck radiation treatment planning.