1) To examine trends in the use of diabetes medications and 2) to determine whether physicians individualize diabetes treatment as recommended by the American Diabetes Association (ADA).We conducted a retrospective, cross-sectional analysis of 2003-2016 National Health and Nutrition Examination Survey (NHANES) data. We included people ≥18 years who had ever been told they had diabetes, had an HbA1C >6.4%, or had a fasting plasma glucose >125 mg/dL. Pregnant women, and those aged <20 years receiving only insulin were excluded. We assessed trends in use of ADA's seven preferred classes from 2003-2004 to 2015-2016. We also examined use by hypoglycemia risk (sulfonylureas, insulin, and meglitinides), weight effect (sulfonylureas, thiazolidinediones [TZDs], insulin, and meglitinides), cardiovascular benefit (canagliflozin, empagliflozin, and liraglutide), and cost (brand-name medications and insulin analogs).The final sample included 6,323 patients. The proportion taking any medication increased from 58% in 2003-2004 to 67% in 2015-2016 (P < 0.001). Use of metformin and insulin analogs increased, while use of sulfonylureas, TZDs, and human insulin decreased. Following the 2012 ADA recommendation, the choice of drug did not vary significantly by older age, weight, or presence of cardiovascular disease. Patients with low HbA1C, or HbA1C <6%, and age ≥65 years were less likely to receive hypoglycemia-inducing medications, while older patients with comorbidities were more likely. Insurance, but not income, was associated with the use of higher-cost medications.Following ADA recommendations, the use of metformin increased, but physicians generally did not individualize treatment according to patients' characteristics. Substantial opportunities exist to improve pharmacologic management of diabetes.