Purpose. To investigate the relationship between size of demyelinated lesion, extent of axonal loss, and degree of latency delay of visual evoked potentials (VEPs) in a rat model of experimental demyelination. Methods. Lysolecithin 1% (0.4 or 0.8 μL) was microinjected into an optic nerve of each of 14 rats 2 mm posterior to the globe. Standard flash VEPs were recorded with skull-implanted electrodes before and 2, 4, and 6 days after the microinjection. The optic nerves were stained with Luxol-fast blue and Bielschowsky's silver to assess demyelination and axonal pathology, respectively. Demyelinated areas were measured on serial sections, and lesion volumes were deduced by three-dimensional reconstruction. Results. Focal lesions of demyelination and variable axonal loss were observed. The mean volume of the lesion was 3.2 ± 1.1 × 10-2 mm3. The injected eye showed a significant latency delay and amplitude decrease. Regression analysis demonstrated a strong correlation between N1 latency delay and lesion volume (r = 0.863, P < 0.0001), which remained significant after adjustment for axonal loss (r = 0.829, P < 0.001). N1 latency delay also showed a correlation with axonal loss (r = 0.552, P = 0.041), but the correlation became nonsignificant when controlling for demyelination (r = 0.387, P = 0.191). A linear association between N1-P2 amplitude decrease and axonal loss (r = 0.681, P = 0.007) was also observed. Conclusions. The latency of the VEP accurately reflected the amount of demyelination in the visual pathway, whereas the amplitude correlated with axonal damage. This study supports the concept that the VEP provides a highly sensitive tool with which to measure demyelination in optic neuritis.