The spin-lattice relaxation time T1 of multicomponent tissues is often determined by fitting relaxation data to monoexponential functions. This process can lead to large errors in the relaxation time. We describe a procedure using chemical-shift imaging (CSI) which separates the NMR signal into water and lipid components thus allowing the two signals to be individually analyzed for relaxation times. This procedure yields more representative relaxation times than those obtained by both monoexponential and biexponential fitting schemes.