Sixteen normal subjects and three patients with optic neuritis were studied to determine the effect of decreased retinal illumination on simultaneously recorded pattern electroretinograms (PERG) and visual-evoked potentials (VEP). Using neutral-density filters (NDF), it was found that linear modeling is an excellent fit for VEP/PERG amplitudes and latencies as log functions of retinal illumination, both for individual eyes and averages of pooled data. Within narrow statistical limits, regression slopes show that mean PERG B-wave and VEP P100 latencies are affected almost identically by decreased illumination, leaving the mean retinocortical time (RCT) virtually unchanged. However, mean B-wave amplitude was greatly reduced at retinal illuminations at which P100 amplitude was unaffected. Of clinical significance was that these latency and amplitude effects were found in each eye tested, whether normal or pathologic. In particular, the RCT in normal subjects was never found to be statistically abnormal due to decreased retinal illumination, and it faithfully represented the optic nerve lesion in the patients with optic neuritis. This result was applied to a population of eight patients with uncomplicated cataracts. The significance of these results is discussed.