OPV2V is the first large-scale public V2V (Vehicle-to-Vehicle) collaborative perception simulation dataset, which is mainly collected by the most popular autonomous driving simulator CARLA and the collaborative driving simulation framework OpenCDA. The dataset has 73 different scenes, including 11,464 frames and more than 230,000 annotated 3D detection frames. The dataset covers 6 types of roads and 9 different cities, 8 of which are cities that come with CARLA, and 1 is a digital city with real traffic flow that is nearly 100% restored based on Culver City, Los Angeles.