PURPOSE: To see the stability and error in the diffusion-weighted magnetic resonance (MR) imaging technique in the experimental models and to observe the signal intensities in the early cerebral lesions of the animal models. MATERIALS AND METHODS: Diffusion coefficients of acetone and distilled water were measured by diffusion-weighted MR image and were compared with actual values. Differentiation of diffusion from perfusion were done at the resin flow phantom. The signal intensities caused by early parenchymal changes were measured in normal, hypovolemic, and embolic, and dead animal models by using diffusion-weighted image and compared with pathoIogic finding and vital staining. RESULTS: Diffusion coefficients of acetone and distilled water were 4.48 x 10-3 and 2.72 x 10-3 which were very close to the actual values. Diffusion-weighted MR image obtained at flow phantom was not affected by flow (perfusion) at the 100-400 of b-factor range. Animal study done at that b-factor range revealed a significant signal difference between the left and right sides only at the embolic model induced by polyvinyl alchol particles (p<0.05). These changes were not detected in microscopic finding but could be identified in vital staining. CONCLUSION: Diffusion-weighted MR image can be used to detect early parenchymal change when the appropriate b-factor range was applied.