Objective To identify plasma miRNAs that are related to treatment failure in youth with type 2 diabetes (T2D). Research Design & Methods Using data and samples from the Treatment Options for Diabetes in Youth (TODAY) Study (n=209), we examined whether miRNAs could predict treatment failure over an average of 962 days of follow-up using glmnet. We also examined whether individual miRNAs were associated with higher risk of failure independent of baseline clinical factors (hemoglobin A1c, Tanner stage, race, BMI, and sex). Results Participants were 14.5 years old, 62% female. 45% (n=94) had treatment failure. Plasma miRNAs alone did not predict treatment failure. Higher A1c was associated with higher risk of treatment failure (HR for 1-unit increase 1.35, p<0.002), and White race with a lower risk of failure compared to Black race (HR -0.68, p=0.05). Two miRs were associated with failure independent of clinical factors. miR-4306 was associated with a 15% decrease in failure for each doubling in expression. miR-483-3p was associated with a 15% increase in failure for each doubling in expression. The addition of miR-483-3p and miR-4306 improved fit in a multivariate model of treatment failure (log likelihood without miRNAs -420.9 vs. with miRNAs -413.3, p<0.001). Identified miRNA were predicted to have roles in retinoic acid receptor signaling, cardiac hypertrophy signaling, serotonin receptor signaling, and insulin secretion signaling. Conclusions A panel of plasma miRNAs alone did not predict treatment failure in TODAY participants. Higher levels of miR-483-3p and lower levels of miR-4306 are associated with treatment failure in youth with T2D treated with metformin, rosiglitazone, and/or lifestyle modification.