We sought to determine use of any and at least moderate-intensity statin therapy in a national sample of patients with diabetes mellitus (DM), with the hypothesis that nationwide frequency and facility-level variation in statin therapy are suboptimal. We sampled patients with DM age 40 to 75 years receiving primary care between October 1, 2012, and September 30, 2013, at 130 parent facilities and associated community-based outpatient clinics in the Veterans Affairs Health Care System. We examined frequency and facility-level variation in use of any or at least moderate-intensity statin therapy (mean daily dose associated with ≥30% low-density lipoprotein cholesterol lowering). In 911 444 patients with DM, 68.3% and 58.4% were receiving any and moderate- to high-intensity statin therapy, respectively. Patients receiving statin had higher burden of cardiovascular disease, were more likely to be on nonstatin lipid-lowering therapy and to receive care at a teaching facility, and had more frequent primary-care visits. Median facility-level uses of any and at least moderate-intensity statin therapy were 68.7% (interquartile range, 65.9%-70.8%) and 58.6% (interquartile range, 55.8%-61.4%), respectively. After adjusting for several patient-related and some facility-related characteristics, the median rate ratios for any and moderate- to high-intensity statin therapy were 1.20 (95% confidence interval: 1.18-1.22) and 1.29 (95% confidence interval: 1.24-1.33) respectively, indicating 20% to 29% variation in statin use between 2 identical patients receiving care at 2 random facilities. Statin use was suboptimal in a national sample of patients with DM with modest facility-level variation, likely indicating differences in statin-prescribing patterns.