PURPOSE: Afferent asymmetry of visual function is detectable in both normal and pathologic conditions. With a computerized test, we assessed the variability in measuring afferent asymmetry of the pupillary light reflex, that is, the relative afferent pupillary defect. METHODS: In ten normal subjects, pupillary responses to an alternating light stimulus were recorded with computerized infrared pupillography. The relative afferent pupillary defect for each test was determined by using a new computer analysis. The 95% confidence interval of each determination of relative afferent pupillary defect was used to represent the short-term fluctuation in its measurement. To optimize the test for clinical use, we studied the influence of stimulus intensity, duration, and number on the variability of the relative afferent pupillary defect. RESULTS: When the relative afferent pupillary defect was based on only a few light alternations (stimulus pairs), there was excessive variability in its measurement (95% confidence interval > 0.5 log units). With approximately 200 stimulus pairs, the 95% confidence interval was reduced to less than 0.1 log unit (relative afferent pupillary defect +/- 0.05 log unit). Also, there was less variability when the dark interval between alternating light stimulation was less than one second. CONCLUSIONS: Computerized infrared pupillography can standardize the alternating light test and minimize the error in quantifying a relative afferent pupillary defect. A reproducible relative afferent pupillary defect measurement is desirable for defining afferent injury and following the course of disease.