Purpose: Evidence shows that tight glycemic control mitigates the adverse microvascular and macrovascular effects of diabetes. Metformin, an oral anti-hyperglycemic agent for type 2 diabetes, is commonly prescribed as first-line treatment for type 2 diabetes. Despite metformin's demonstrated efficacy, low rates of treatment persistence (the proportion of patients remaining on medication for a period of time) are common among diabetes patients. We examined persistence rates associated with immediate-release metformin (MIR) versus a novel extended-release metformin (MER) formulation designed to improve tolerability and efficacy. Methods: The study was a phase III, 24-week, randomized, double-blind, active-controlled, fixed-dose trial comparing MIR 1500 mg/day, b.i.d. (MIR-1500B) versus MER at three doses [1,500 mg/day q.d. (MER-1500Q), 1,500 mg/day b.i.d. (MER-1500B), or 2,000 mg/day q.d. (MER-2000Q)]. Participants were adults with type 2 diabetes who were medication naïve or received prior oral hypoglycemic monotherapy. Titration to study dose was achieved over 3 weeks. Analyses were conducted using a modified intent-to-treat (ITT) approach, whereby participants who completed the study titration phase, received one week of study dose, and had at least one A1C follow-up from baseline were selected. Reasons for study withdrawal (initiated by investigator or patient) included lack of efficacy, serious or non-serious adverse events, death, patient desire to withdraw, patient noncompliance with protocol, and lost to follow-up. Patients who prematurely terminated for any reason were defined as non-persistent; those who were titrated to the full study dose and completed the study through week 24 were considered to be persistent. Results: The ITT analysis included 647 participants with a mean baseline A1C of 8.3% (SD 1.5). Treatment groups had similar demographics and disease-related characteristics at baseline. Patients who received MER-2000Q were approximately half as likely to prematurely terminate from the study as those receiving MIR-1500B (Odds Ratio: 0.52, 95% CI 0.28 to 0.96, p=0.04). Patients receiving MER-2000Q had a lower rate of withdrawal due to lack of efficacy (1.8%) compared with those receiving MIR-1500B (9.9%, NS) and MER-1500Q (11.4%, p=0.03). Groups were similar with respect to the number, type, or severity of adverse events. Conclusions: Higher treatment persistence is associated with better glycemic control. Persistence rates were two times higher with extended-release versus immediate-release metformin, possibly due to enhanced tolerability and/or efficacy associated with the maximum Glumetza 2000 mg QD dose.