AIM: To estimate the adult reference values for measured deviations by a computerized diplopia test and testify the validity. METHODS: Totally 391 participants were recruited and taken the computerized diplopia test. The plots and amplitude of deviations were recorded. The differences in different gender, age and visual acuity groups were analyzed respectively. Of 30 subjects were enrolled to testify the interobserver reliability. Another 46 subjects (including 26 normal subjects and 20 patients) were taken the test and theirs deviations were recorded to testify the validity of the reference value. RESULTS: The max horizontal and vertical deviations were 2.55° and 0.76° with normal corrected visual acuity while 3.88° and 1.46° for subjects with poor corrected vision. The differences between age groups was insignificant (Z=3.615, 4.758; P=0.461, 0.313 for horizontal and vertical respectively). The max horizontal deviation of female was smaller than male (Z=-2.177; P=0.029)，but the difference in max vertical deviation was insignificant (Z=-1.296; P=0.195). The mean difference between observers were both -0.1°, with 95% confidence limits (CI) of -1.4° and 1.6° in max horizontal deviations while -2.1° and 1.8° in max vertical deviation. The mean deviation of 26 normal subjects was 1.02°±0.84° for horizontal and 0.47°±0.30° for vertical which both within the range of reference values. The mean deviation of 20 patients was 13.51°±11.69° for horizontal and 8.34°±8.58° for vertical which both beyond the reference range. CONCLUSION: The max amplitude of horizontal and vertical deviation is pointed as the numerical parameters of computerized diplopia test. The reference values are different between normal corrected visual acuity and poor corrected vision. These values may useful for evaluating patients with diplopia in veriety conditions during clinical practice.