Abstract: Rivers cover only 0.58% of the earth's non-glacial land surface, but they play a vital role in the global carbon cycle. In the past, there were many blind spots in the study of carbon dioxide (CO₂) emissions at the water-air interface of global rivers, resulting in great uncertainty in the estimation of global river CO₂ emission fluxes. In recent years, with the improvement of measurement methods and the accumulation of global data, many studies worldwide have revised and updated the fluxes. This paper summarizes the spatial and temporal variability of global river partial pressure of CO₂ (pCO₂) and CO₂ emission fluxes as well as the related influencing factors from the temporal and spatial perspectives, respectively. It further assesses the impacts of climate change and anthropogenic perturbations on river carbon fluxes. The results demonstrate that the pCO₂ in the surface waters of rivers and streams exhibits significant spatial and temporal heterogeneity, and it decreases from low latitudes to high latitudes, while pCO₂ in tropical rivers and streams is generally higher than that of temperate, northern and Arctic regions. There are obvious seasonal differences of pCO₂ values in the surface waters of rivers and streams with a generally higher value in summer than in winter. At present, the latest estimate of the annual CO₂ emissions from global streams and rivers is (2.0 ± 0.2) Pg/a, and CO₂ flux from streams and rivers gradually decreases with the increase of river branch grade. The role of small rivers in carbon emissions within river ecosystems requires attention. The rivers and streams situated within distinct geographic contexts possess unique mechanisms of CO₂ production and emission. Climate change and human disturbances (such as dam construction, urbanization, agricultural irrigation, etc.) can increase the uncertainty of CO₂ flux estimation from rivers. Furthermore, the estimation of CO₂ emission fluxes at the water-gas interface of rivers and streams across the globe continues to be subject to spatial and temporal biases. These biases necessitate further quantification and refinement at a more granular river network scale to enhance the accuracy of CO₂ flux assessments.