Stable carbon isotopes (δ13C) are a useful technique to partition the net ecosystem–atmosphere exchange (NEE) of carbon dioxide (CO2) into gross primary productivity (GPP) and respiratory (R) components. The acquisition of ecosystem flux is usually performed using the eddy covariance (EC) technique, but this method is limited under complex terrain conditions and stable atmospheric conditions. Here, we try to develop the flux partitioning theory of mountain forest ecosystem by combining isotope technology and soil respiration. We used a full growing season of isotopic profile data and soil respiration data from a temperate deciduous forest on Jiufeng Mountain to demonstrate the method and quantify its uncertainties. We found that the assessable systematic uncertainty was ± 29.97% for NEE, ±24.87% for GPP, and ± 36.82% for R. The factors that most strongly affected the uncertainties of GPP and R were canopy CO2 stomatal conductance (gco2), calculated by the Köstner model, and the carbon isotope value of the total respiration in the ecosystem at night (δ13Cn-eco), calculated by the Keeling plot method. The advantage of isotopic methods is not only the relief of terrain restrictions but also the stable atmospheric conditions. The method provides an innovative vision for the partitioning of flux in forest ecosystems, especially for mountain forests. © 2022