ResOps-BR is a hydrometeorological dataset covering 142 reservoirs in Brazil. The dataset includes, apart from time series of reservoir operations downloaded from the Agência Nacional de Águas, hydrometeorological time series, and catchment, reservoir and dam characteristics. The final purpose is to train hydrological models that reproduce reservoir operations. It has been developed for comparing reservoir models of different nature —process-based and data-driven—. It is precisely the idea of applying data-driven models, such as deep learning, that required the addition of static attributes (reservoir, dam and catchment characteristics). To ease the use of already existing deep-learning frameworks, the dataset structure is identical of that of the CARAVAN initiative. We put special focus in developing a dataset framework that can be applied to any other country. For that reason, we use open datasets like GDW (Global Dam Watch) for the reservoir characteristics, CEMS GloFAS (Copernicus Emergency Management Service Global Flood Awareness System) for the hydrological time series or ERA5 for the meteorological time series. ResOpsBR+CARS is the second of an array of national reservoir operations datasets that will build a global reservoir operations dataset that could be used to improve reservoir representation in large-scale hydrological models. The previous dataset covered the USA: ResOpsUS+CARS.