A retrospective analysis was performed on the transient and steady-state pattern electroretinograms recorded from 42 patients with glaucoma, 13 patients with senile dementia of the Alzheimer's type, 58 patients with diabetes mellitus, and 92 control subjects to evaluate the pattern of electroretinographic changes associated with retinal and optic nerve disease. The amplitudes of both the initial positive component (N1 to P1) and the subsequent negative component (P1 to N2) of the transient (4 rps) responses were measured. From these measurements the (P1 to N2)/(N1 to P1) was derived. The N1 to P1 amplitude of the steady-state pattern electroretinogram also was measured. In the glaucoma patients all three amplitude measures, as well as the amplitude ratio of the components of the transient response, were reduced significantly compared with age-matched controls (p less than 0.05). A similar pattern was detected in the patients with Alzheimer's disease, but in this case the only statistically significant amplitude reduction was in the steady-state pattern electroretinogram. A different pattern was observed among the diabetic patients (both with and without retinopathy). Only minor reductions in the amplitude of the transient pattern electroretinogram, which were not statistically significant, were noted. In addition, the ratio of the amplitudes of the components of the transient response did not differ from age-matched controls. The amplitude of the steady-state pattern electroretinogram was reduced in diabetics, but this was significant only for those patients with retinopathy (p less than 0.01). These findings support the suggestion that an analysis of both the positive and negative components of the pattern electroretinogram may be useful for differentiating the contributions of retinal and optic nerve dysfunction to visual impairment. The results also indicate that in both retinal and optic nerve disease the steady-state pattern electroretinogram can be an earlier sign of dysfunction than the transient pattern electroretinogram.