OBJECTIVES: Adherence to anti-diabetic medications has always been a concern due to its importance in diabetes management. Adherence to medication can depend on the patients’ compliant behavior and the patients’ perception of the severity of their diabetes. This study aims to assess the effect of past compliant behavior and severity of diabetes on patients' adherence to their anti-diabetic medications. METHODS:  This is a retrospective database study of continuously enrolled (2009-2010) Texas Medicare Advantage Plan enrollees (≥60 years) with ICD-9 diagnosis codes for diabetic disorders. Patients were followed for one and a half year to measure their adherence to antidiabetic medication. Prescription claims for biguanides, diuretics, and sulfonylureas were utilized to determine medication adherence using medication possession ratio (MPR). The MPR 365 days following index date was used as a proxy measure of past compliance. Severity was assessed based on the presence of diabetic neuropathy (DN) in first 365 days of follow-up, with patients having DN being categorized as severe. The impact of patients' past compliant behavior and the presence of DN was assessed on the adherence during the 180 days after the compliance measurement period using multivariable logistic regression analysis. RESULTS: A total of 1,635 diabetic patients were identified with mean age 73.3(+\-7) years of which 58.9% were female. Almost 10% had diabetic neuropathy, and 66.5% patients were adherent to their medication in first 365 days of follow-up. Logistic regression showed that patients with DN were more likely to be adherent compared to the patients without DN (OR=3.46, 95% CI [1.91-6.26]); similarly, patients who were compliant to their medication in past were more likely to be adherent to their medications (OR=1.45, 95% CI [1.10-1.92]). CONCLUSIONS: Severity of diabetes (presence of DN) and previous compliant behavior is positively associated with the adherence to anti-diabetic medications.