There are no data in children with type 2 diabetes (T2D) regarding the durability of glycemic control with oral medication. Therefore, we assessed the likelihood of and time to failure of oral therapy in children and adolescents diagnosed with T2D. Charts of patients presenting to our large tertiary-care children's hospital between January 2000 and June 2007 with a new diagnosis of diabetes (n = 1625) were reviewed to identify those with T2D (n = 184). Subjects' initial therapy, hemoglobin A1c (HbA1c), body mass index, age, gender, and antibody status were documented, as well as subsequent therapies and HbA1c values, to determine whether baseline characteristics predicted future insulin dependence. Kaplan–Meier survival curves and Cox proportional hazards analysis demonstrated time to failure of oral therapy. Eighty-nine patients remained on insulin throughout the study. Baseline characteristics that determined future insulin dependence included being placed on insulin initially, initial HbA1c and race (whites less likely to be insulin dependent at study conclusion). Patients who failed oral therapy were more often reported to be non-compliant or unable to tolerate metformin than those who continued on oral therapy. The median time to failure of oral therapy (metformin monotherapy in 84/95) was not significantly different for patients initially treated with oral therapy (42 months) and insulin (35 months). Thus, children with T2D appear to fail oral therapy more quickly than what is reported in adults. It is not yet known if improving compliance with treatment might allow more children to remain on oral medications.