OBJECTIVES: To compare the main outcomes differences including clinical events, health care utilization and costs of patients using an insulin device for diabetes versus patients using the conventional human insulin vial/syringe. METHODS: Using a retrospective analysis of health insurance claims data between the years 2003 and 2008, we identified patients with a diagnosis of diabetes and then divided them into an insulin device cohort and a human insulin vial/syringe cohort, based on their prescription fills. Patients’ demographics, health care visits and costs were compared using Chi-square testing and standardized differences. The 12-month follow-up clinical event rates, health care facility use and costs for those patients were compared. Risk adjustment was performed using the propensity score matching method with the ProbChoice™ algorithm. RESULTS: A total of 12,400 eligible patients were identified as using insulin for diabetes: 1,236 (9.97%) received the insulin device and 11,164 (90.03%) received the insulin vial/syringe. Compared with patients who received the conventional human insulin vial/syringe, patients in the insulin device group were likely to be younger, live in the Midwest of the United States, and have type I diabetes. Although there were no significant differences in hypoglycemic events after risk adjustment, patients in the insulin device group had significantly fewer cases of cerebrovascular disease (4.14% vs. 9.12% p=0.0055), congestive heart failure (7.18% vs. 12.15% p=0.0267) and chronic obstructive pulmonary disease (4.70% vs. 10.50% p=0.0039), but more cases of  dyslipidemia (68.51% vs. 54.42% p=0.0002). Although the outpatient costs for office visits ($1888 vs. $1895 p=0.0257) were lower for patients on the insulin device, their prescription costs ($5489 vs. $4635 p<0.0001) were higher. The overall risk-adjusted healthcare costs did not differ ($14,231 vs. $18,096 p=0.1160) between the two groups. CONCLUSIONS: Without significant addition to the costs, insulin administration with the device is associated with fewer clinical events.