OBJECTIVE: To examine longitudinal outcomes, rate of and reasons for discontinuation, and predictors of insulin pump success in a cohort of youth initiating pump therapy. RESEARCH DESIGN AND METHODS: We followed a cohort of youth with type 1 diabetes (n = 161) starting the pump between 1998 and 2001 and recorded natural history of treatment. RESULTS: At pump start, patients (71% female) had a mean age of 14.1 +/- 3.7 years, diabetes duration of 7.1 +/- 4.0 years, daily blood glucose monitoring (BGM) frequency of 4.0 +/- 1.2, a daily insulin dose of 1.0 +/- 0.3 units/kg, and an HbA(1c) (A1C) of 8.4 +/- 1.4%. After 1 year, mean daily BGM frequency was 4.5 +/- 1.7, daily insulin dose was 0.8 +/- 0.2 units/kg, and A1C was 8.1 +/- 1.3% (all baseline versus 1-year data, P < 0.01). As of 2005, 29 patients (18%) had resumed injection therapy at a mean age of 17.0 +/- 2.9 years after a mean duration of pump use of 2.1 +/- 1.3 years. BGM frequency at baseline and at 1 year was significantly lower in the patients who resumed injection therapy (P < 0.02). In addition, patients who remained on the pump had lower A1C than those who resumed injection therapy at both 1 year (P = 0.04) and at the most recent clinic visit (P = 0.01). CONCLUSIONS: After an average of 3.8 years, >80% of pediatric patients maintained pump therapy with preservation of baseline A1C. Patients discontinuing the pump were less adherent and did not achieve equivalent glycemic benefit compared with continued users; these patients require ongoing support aimed at improving adherence and outcomes.