We studied 45 primary open-angle glaucoma patients consecutively examined with the program 30-2 on the Humphrey Field Analyzer to determine factors associated with a learning effect between the first two visual field exams. The average mean defect significantly decreased 1.7 dB between exams (P less than 0.05). No difference was noted in learning based on: the time interval between tests, the area of the field tested, age, sex, race, previous perimetric experience or parameters on the initial field exam (i.e. prevalence of false positive and negative errors, central reference level and eye tested (P greater than 0.05). However, learning positively correlated to defect depth and negatively correlated to fixation losses on the initial exam in patients retested within seven months (P less than 0.05). This study suggests predicting which patients require a second baseline field exam to exclude a learning effect is difficult, except in those with glaucomatous field loss or few fixation losses and another field is planned within a half year.