Few studies have focused on metabolite diffusion using 1H-MRS, compared to the vast number observing water diffusion by DWI/DTI. These MRS studies are lengthy, therefore difficult to implement clinically, and use up to three b-values to yield the diffusion spectra. Single exponential signal loss is assumed for metabolites, neglecting the possibility of non-linear decay at high b-values, as has been observed for water. Our goals are: (i) Characterize the metabolite signal decay versus b-value in human white matter to determine the non-linear region. (ii) Develop a rapid diffusion tensor spectroscopy method that can be executed in a clinically useful time.