We assessed the influence of frequency variation on measurements of fatty acid composition in adipose tissue. A 3D bilateral gradient spectroscopic imaging sequence with a simultaneous dual-slab excitation followed by 128 monopolar echoes was used. The sequence included a short train of 12 navigator echoes without phase encoding at the beginning of each TR period to correct frequency variations due to hardware heating and patient respiration. The proposed method was tested in oil phantoms and ten postmenopausal women. Phase correction reduced measurement error in the phantom and spatial variation in estimates of fatty acid composition in vivo.