The Schirmer test was performed 1350 times in 110 normal individuals and 15 dry eye patients to investigate the significance of the test as a diagnostic method for dry eye. The reproducibility of the Schirmer test was 54.5% in normal individuals and 41.9% in dry eye patients. There was no difference in the reproducibility between the groups with topical anesthesia and those without it. The ratio of misdiagnosis by Schirmer test was 48.4%. These results suggest that it is impossible to differentiate dry eye patients from normal individuals by the Schirmer test.