We evaluated the reliability of optic disk area and disk rim area measurements obtained with the Humphrey Retinal Analyzer from ten repeat fundus images each of one eye of ten normal subjects and ten subjects with increased intraocular pressure. Variability of rim and disk areas was not significantly different between these two groups (P greater than .05), nor was it significantly different between images acquired during the same session and those acquired during different sessions (P greater than .05). Variability was significantly different between subject groups based on the distinctness of the disk edge and the slope of the cup wall (P less than .01). The average minimum change in the measurements on the same eye that would exceed variability of the technique at the 95% confidence level was 0.20 mm2 and 0.47 mm2 for rim area for the least variable and most variable subject groups based on optic nerve head characteristics, and 0.09 mm2 and 0.32 mm2 for disk area for the corresponding subject groups. Operator input contributed significantly to overall variability. Reliability of rim and disk areas in this study was comparable to that reported for other computerized image analyzers.