The aim of this study was to investigate the test-retest reliability of the Arden Grating Test (AGT), and to assess the extent to which any variance in AGT score on retest can be attributed to intra-subject and inter-tester differences. Twelve patients with various pathologies and whose contrast sensitivity covered a wide range were each tested twice by six testers. It was shown that variance attributable to testing by different clinicians accounted for approximately 25% of the total variance, and that the 95% confidence limits about any observed score were +/- 15 AGT units, which is about one quarter of the total dynamic range of the test. The large range of scores encompassed by the 95% confidence limits represents considerable unreliability, the consequence of which is a high misclassification rate, i.e. many false positives and false negatives, especially in the context of primary vision screening for which it was designed.