OBJECTIVES: Glucagon-like peptide-1 receptor agonists (GLP-1RAs), such as exenatide once weekly (EQW) and liraglutide (LIRA), are FDA-approved as treatment for type 2 diabetes mellitus (T2DM).  Head-to-head studies and meta-analyses of these agents have reached different conclusions about their relative effectiveness.  METHODS: We developed a decision-analytic model to evaluate the likely incremental cost-effectiveness of EQW versus LIRA 1.8 mg in T2DM patients, with effectiveness measured as reduction in glycated hemoglobin (HbA1c). The model structure tracks change in HbA1c and direct medical expenditure (drug, needle, adverse events [AEs]) over a 26-week time horizon, and allows patients to discontinue treatment due to AEs (nausea, diarrhea, vomiting, constipation, dyspepsia) after 1 or 3 months of therapy. Patients discontinuing treatment are assumed to return to their baseline HbA1c. We compared the outcomes (cost per 1% and 0.2% reduction in HbA1c) of models populated with clinical data from a head-to-head 26-week randomized, controlled trial (DURATION-6) and a meta-analysis conducted by Scott and colleagues (2012).  Drug costs and other utilization costs were based on wholesale acquisition costs and published sources.  RESULTS: For the base case, the projected total 6-month cost of EQW versus LIRA was $2,444 and $3,054, respectively.  Using data from DURATION-6 and meta-analysis, compared with EQW, LIRA had a projected incremental cost per 1% reduction in HbA1c (ICER) of $3,262 and $18,578 over a 6-month time horizon, respectively.  Compared with EQW, the projected 6-month cost per 0.2% reduction in HbA1c for LIRA was $652 and $3,716 based on data from DURATION-6 and meta-analysis, respectively.  CONCLUSIONS: The projected cost per 1% reduction in HbA1c was lower with EQW than liraglutide 1.8 mg at 6 months.  The difference in projected cost per HbA1c reduction varies significantly depending on the trial-based data sources used.  Real-world data are needed to resolve the current uncertainties.