Abstract The Amazon basin has been subjected to extreme climatic events and according to climate change projections this hydrosystem could face changes in the natural dynamic of flood cycles that support the feeding and reproduction of many fish species, threatening aquatic biodiversity. Protected areas (PAs) are the main tools used to safeguard the biodiversity in the long term; however, they are fixed areas that could be subject to climate change, questioning their future efficiency in protecting biodiversity. The Amazon basin currently benefits from a relatively high level of protection as 52% of its catchment area is under the form of true PAs or indigenous lands. However, the capacity of these PAs to protect freshwater biodiversity remains unclear as they have generally been assessed with little regard to freshwater ecosystems and their hydrological connectivity. Here, the aim was to evaluate the effectiveness of PAs in representing the Amazon fish fauna under current and future climatic conditions. A macroecological approach was used to estimate the minimum size of the geographical range needed by each species to achieve long‐term persistence, by a combined function of range size and body size, two ecological traits known to influence species extinction risk. In future the Amazon basin could risk losing 2% of its freshwater fish fauna owing to unsuitable climatic conditions, with a further 34% adversely affected. The present Amazon network of PAs will cover the minimum required range for species persistence for more than 60% of the freshwater fish species analysed under the future climate scenario. However, more than 25% of the future susceptible species are currently concentrated in large tributaries and in the central‐lower Amazon floodplain where few PAs occur, highlighting the lack of appropriate conservation actions for these specific water bodies.