BACKGROUND In the Erlangen glaucoma study, the blue-on-yellow VEP was shown to be able to discriminate between controls and manifest glaucoma patients. SCIENTIFIC QUESTIONS AND AIMS: In our investigation, we assessed the validity of the blue-on-yellow VEP for early diagnosis of glaucoma. With this aim, we compared different subgroups of glaucoma suspects. The main issue of the investigation was the biostatistical aspects of early diagnosis of glaucoma. MATERIAL, METHODS AND RESULTS Within a group of patients who were suspected of having ocular hypertensive glaucoma without visual field loss we compared 109 patients with optic disc damage [preperimetric (PPM) 47 +/- 11 years] and 91 patients without optic disc damage [ocular hypertension (OHT) 45 +/- 10 years]. We evaluated the N 1-amplitude and the peak latency of the blue-on-yellow VEP. The peak latency was significantly longer in the PPM group (first examination: OHT 118 +/- 9.5 ms, PPM 122.0 +/- 10.5 ms; second examination: 119.1 +/- 7.4/121.9 +/- 11.0 ms; third examination: 118.5 +/- 9.1/122.4 +/- 10.9 ms). The amplitude was reduced in the PPM group (P = 0.08). The differences between the two groups only allowed limited individual separation: (sensitivity of 42% for advanced optic disc damage with 80% specificity among OHT patients). CONCLUSIONS The reduced sensitivity of a diagnostic procedure within a group of glaucoma suspect patients compared with patients with manifest glaucoma might be explained by: (1) possible misclassifications of patients and (2) a smaller degree of loss of visual function in the early stages of the disease.