Objective: We report on the association between traditional Chinese medicine (TCM) constitution and diabetes mellitus (DM) and provide epidemiological evidence for the theory of correlation between constitution and disease. Methods: A total of 3748 participants were used for data analysis, and all study subjects underwent a complete clinical baseline characteristics evaluation to collect related information. DM was determined by an oral glucose tolerance test. Multiple logistic regression (MLR) models were employed to detect the associations. Results: In total, 16.84% of the participants had DM in the total sample. Significant differences were found in age, height, weight, and heart rate, between the DM and non-DM groups. Univariate linear regression analyses indicated that the variables Qi_Deficient and Yang_Deficient were significantly associated with the outcome, and a negative correlation between TCM constitution and DM was found. After adjustment for relevant potential confounding factors, the MLR detected significant associations between both the Qi_Deficient and Yang_Deficient variables and DM ([Formula: see text]-value [Formula: see text] 0.015 for Qi_Deficient and [Formula: see text]-value [Formula: see text] for Yang_Deficient). In Qi_Deficient and Yin_Deficient participants, the odds ratios (ORs) for DM were 0.778 and 0.646, respectively. Conclusion: Qi_Deficient and Yang_Deficient were significantly associated with DM. These findings may provide insights for clinical practice for the prevention and diagnosis of DM.