To detect the color discrimination changes in early-stage diabetic patients. Retrospective series case study. A total of 78 type 2 diabetic patients (126 eyes) without diabetic retinopathy were enrolled. Another 64 healthy individuals were selected as the normal control. Total error scores (TES) and each error score (ES) in distinguishing the red, green and blue colors were measured via the Farnsworth-Munsell 100 hue tester. The TES origin scores and their square roots were used for statistical analysis. The results of the two groups were compared. The influence of gender, age, duration of diabetic mellitus and application time of insulin on TES and each color's ES values was investigated. The t, t' or u-test was adopted to compare the mean of TES and ES values between these two groups; the Spearman rank correlation analysis was used to analyze the relationship between TES and ES values and age, duration and duration of insulin usage; the t test was employed to analyze the relationship between TES and ES values vs. gender; the number of superior, average, low discrimination eyes were analyzed via the χ(2) test. Wilcoxon rank sum analysis was adopted in determining the color discrimination ability between these three groups. The TES and its square root values in the diabetic and normal groups were 63.80 ± 34.19, 30.39 ± 4.57 and 7.69 ± 2.17, 5.50 ± 0.42, respectively. There were differences in both values between these two groups (t = 4.87, 4.91; P < 0.05). The error scores and its square root values in distinguishing the red, green, blue colors in the diabetic group were 9.39 ± 8.61, 12.67 ± 9.71, 12.39 ± 8.05 and 2.62 ± 1.60, 3.23 ± 1.51, 3.28 ± 1.27, respectively, and the values were 5.13 ± 3.59, 7.00 ± 1.84, 6.81 ± 2.70 and 1.95 ± 1.17, 2.62 ± 0.36, 2.52 ± 0.68, respectively, in the normal group. There were differences between these two groups (u = 3.62, 4.94, 5.40 and 2.40, 3.32, 4.03; P < 0.05). Differences existed in discriminating the three colors between two groups (u = 4.071, P < 0.05). Except for age (r = 0.38, P < 0.05), there was no relationship between color vision defects and age, duration (r = -0.02 ∼ 0.23, P > 0.05), gender (r = 0.32 ∼ 0.39, P > 0.05) or duration of insulin usage (r = -0.03 ∼ 0.11, P > 0.05). Color vision defects appear in early-stage diabetic patients. Color discrimination can play an important role in the evaluation of color perception damage and visual function in diabetic patients.