OBJECTIVES: Evidence suggests depression rates are high among type-2 diabetic populations and untreated depression can complicate effective management of diabetes. However, accurately characterizing depression in research study cohorts can be challenging. As part of the Health Related Quality of Life(HRQL) substudy in the Action to Control Cardiovascular Risk in Diabetes(ACCORD) trial, we collected data on depression from 2053 adult type-2 diabetics in North America. We use baseline data to evaluate the overlap between depression states and associations with selected demographic, lifestyle, and physical variables. METHODS: The 9-item depression scale from the Patient Health Questionnaire(PHQ-9) was used to assess current depression symptoms; history of depression was determined by participant self-report of prior physician diagnosis; and current use of antidepressant medications based on review of medications. Overlap between the three assessments was examined using cross-tabs, and associations with covariates were assessed using univariate logistic regression. RESULTS: Among the 1939 substudy participants with complete data, 701(36.2%) participants had at least one classification of depression, including 537(27.7%) reporting either history(24.6%) or current anti-depressant use(13.9%) and 378(19.5%) a PHQ score higher than 9. Among the latter, 214(56.6%) reported no history of physician-diagnosed depression or use of antidepressant medications. Participants with younger age, female gender, larger BMI, larger waist circumference, higher cholesterol, or insulin use had significantly (p<0.05) increased odds of all classifications of depression. Race/ethnicity, smoking status, HbA1c, serum creatinine, living alone, diet, alcohol use, blood pressure, amputations, QALYs, and other medication use were related to some, but not all definitions of depression. Prior CVD, duration of diabetes, education, GFR, LDL, HDL did not have significant relationships with any definition. CONCLUSIONS: Characterization of depression in research settings is complicated. In ACCORD, we find different covariate relationships for definitions based on history versus current symptoms. These examinations may enhance our understanding of depression's impact on disease management and cardiovascular risk.