Using confocal scanning laser tomography, we studied the test-retest variability of topographic measurements of the optic nerve head and parapapillary retina in 30 patients with glaucoma and 30 normal control subjects. We obtained three independent images, separated by between one and six hours, centered on the optic nerve head. We condensed each original 256 x 256-pixel image to a 64 x 64-pixel image, thereby allowing a realistic estimate of the empiric 90% confidence interval of testretest variability at each condensed pixel. Confidence interval maps generated for each subject showed highest measurement variability along the cup border and blood vessels. The mean standard deviation equivalents of test-retest variability in the patients and controls were 31.20 and 25.94 microns, respectively. These differences were statistically significant (P = .010). Variability increased with age (P < .001). When the analysis was repeated for discrete parapapillary areas, the group differences were not statistically significant (P = .100).