Potential evapotranspiration (PET) is a key factor in calculating actual evapotrans-piration, evaluating regional dry and wet conditions, and water resources management. Using daily meteorological data at 11 typical forest sites of the Chinese Ecosystem Research Network (CERN) in the past 20 years, we estimated PET by both the Penman-Monteith equation and Priestley-Taylor equation, and analyzed the trend of PET and its climate attribution. The results of the two methods indicate that PET in seven typical forest ecosystems in China showed decreasing trends in the past 20 years. Wind speed dominated PET changes in the Changbai Mountain temperate coniferous and broadleaved mixed forest and the He Mountain subtropical artificial evergreen broadleaved forest, while net radiation dominated PET changes in other forests. We also quantified the difference between PET at each forest site (PET_PM) and that estimated by meteorological data from a nearby site of China Meteorological Administration (PET_{_CMA}). On average, PET_{_CMA} is higher than PET_PM, mainly because the meteorological stations are mostly with higher temperature, wind speed, and net radiation, and lower relative humidity than the CERN forest sites. For the forests in northern and eastern China, PET_{_CMA} trends were higher than PET_PM, because the former have higher wind speed and net radiation trends. In contrast, PET_{_CMA} trends were lower than PET_PM for other forests, which are mainly due to their higher trend of relative humidity and lower trend of net radiation. The research can provide a reference for understanding the characteristics of the potential evapotranspiration in China’s forest ecosystems and its response to climate change.