There is limited research using real-world data to evaluate protective cardiovascular effects of glucagon-like peptide-1 (GLP-1) agonists among adults with type 2 diabetes (T2D) early in treatment. We conducted a retrospective, active comparator cohort study using 2011-2015 administrative claims data to compare cardiovascular disease (CVD) event rates following initiation of exenatide extended-release (E-ER), exenatide immediate-release (E-IR) or liraglutide in T2D adults who previously received no other antidiabetic medication (ADM) except metformin. The primary outcome was time to first major adverse CVD event (ischaemic heart disease, stroke, congestive heart failure or peripheral arterial disease) after starting GLP-1. Cox proportional hazards regression was used to model the association between index GLP-1 and CVD events, adjusting for baseline patient, prescriber and plan characteristics. Primary analyses included all patients with ≥2 prescription fills for the index GLP-1, regardless of subsequent refill adherence or initiation of other ADM after index date. Compared with liraglutide, neither E-ER nor E-IR was associated with risk of composite major CVD events (hazard ratios [HRs] for E-ER and E-IR: 1.33 [95% C.I. 0.73-2.39] and 1.30 [0.81-2.09]). No associations were observed between event rates for individual CVD components. The HR for an ischaemic event with E-IR relative to liraglutide was 1.85 (95% C.I. 0.97-3.53). Adjusting for time-varying exposure to other ADM and CVD medications after index date produced similar results. Initiating either immediate or extended-release exenatide rather than liraglutide was not associated with significant differences in CVD risk in this observational real-world study.