Regional and global eddy covariance networks provide the largest comprehensive in situ dataset on carbon fluxes between terrestrial and atmospheric ecosystems. Carbon balance in terrestrial ecosystems can be accounted for bottom-up by these flux data. However, the plausibility of some measurement data has aroused widespread concern from the flux community. This study reviews the research progress over the past four decades, summarizes the principles of eddy covariance measurement, and systematically analyzes the limitations of the instrument and the potential impact of environmental factors. Subsequently, we discuss the controversy over the filling of different carbon flux components, the net carbon uptake phenomenon of ecosystem dormancy in winter against physiological knowledge, the underestimation of ecosystem respiration by insufficient development of nighttime turbulence, and the phenomenon of data offset and measurement lag time due to the above factors. By focusing on the defects and disputes in the measurement principle of eddy covariance and data processing, appropriate improvement measures are proposed to limit (or reduce) the uncertainty of data measurement. This study aims to provide theoretical support for improving monitoring accuracy in the future as well as exploring the carbon cycle of terrestrial ecosystems and its response to environmental factors.