We have previously shown that check-size-specific changes of the Pattern-Electroretinogram (PERG) can be used to detect early glaucoma. This time we conducted a study to determine whether different check sizes, contrasts and eccentricities can improve the diagnostic yield. We examined 15 normal eyes (15 subjects), 32 eyes (24 patients) in stage I or II glaucoma and 36 eyes (25 patients) with ocular hypertension (OHT). Visual acuity was always better than 0.8. The stimuli were checkerboards, phase-reversing at 16 rev/s. The check sizes were 0.8 degrees or 15 degrees with contrast at 60% or 98%. The stimulation area was either "full-field" (27 degrees x 30 degrees), "central" (less than or equal to 7 degrees) or "paracentral" (greater than 7 degrees). In the glaucoma group, all PERG amplitudes were significantly reduced compared to normals: maximally for stimuli of 0.8 degrees, 98% contrast, full-field to 51% +/- 18%, and least for stimuli of 15 degrees, 98% contrast, central (to 70% +/- 20%). Under paracentral stimulation, the amplitude was reduced to 55% and in the central condition to 64%. The single-most-sensitive stimulus to separate normal and glaucoma patients was 0.8 degrees, 60% contrast and full-field, as the variance was low for this condition. Only 4 patients (8.5%) were incorrectly classified using these parameters; when 8 different stimuli were entered into the discriminant analysis, the rate was marginally reduced to 3 patients (6.4%). Seventy-five percent of the OHT eyes were classified as pathological. These results confirm previous findings that in early glaucoma, PERG amplitudes are more reduced using check sizes of 0.8 degrees compared to 15 degrees.(ABSTRACT TRUNCATED AT 250 WORDS)