OBJECTIVES: This study examined different types of medical costs associated with early discontinuation (ED) of insulin therapy and with restarting of insulin among ED patients in the treatment of type 2 diabetes mellitus (T2DM).   METHODS: Truven’s Health Analytics Commercial Claims and Encounters database from 1/1/08 through 12/31/11 were utilized. ED was defined as discontinuation of insulin therapy for at least 30 days in the first 90 days post initiation.  Generalized linear models with log-link were used to examine the association between ED and total medical costs, acute care (hospitalization and emergency room) costs, outpatient costs, diabetes-related drug costs and all-cause drug costs.  Among ED patients, the association between restarting insulin therapy and costs were also examined.  Analyses controlled for patient characteristics, index medication prescribed, patient general health, comorbidities, prior resource utilization, and prior medication utilization.  RESULTS: Most (82% of 74,399) individuals discontinued insulin therapy in the first year post initiation and 75% of these patients were ED.  Compared to non-ED, ED was associated with 9.6% higher acute care costs (P=0.0003), while outpatient costs, diabetes-related drug costs, all-cause drug costs and total medical costs were all significantly lower among ED patients. Among the ED cohort, 90% restarted therapy over the post-period, with 53% restarting within 4 months.  Compared to non-restart, the restart of insulin therapy after ED was associated with 11.3% higher acute care costs (P=0.033), 24.0% higher outpatient costs (P<0.001), 80.2% higher all-cause drug costs (P<0.001, and 30.3% higher total medical costs (P<0.001).  CONCLUSIONS: Current findings suggest that during the first three months following initiation of insulin therapy for T2DM, the ED of insulin therapy and its restart are associated with higher acute care costs. Such costs are often considered avoidable or modifiable costs and tend to signal poorer long-term treatment outcomes.