Abstract: Developing mathematical models for high Knudsen number (Kn) flow for isotopic gas fractionation in tight sedimentary rocks is still challenging. In this study, carbon isotopic reversals (δ13C1 > δ13C2) were found for four Longmaxi shale samples based on gas degassing experiments. Gas in shale with higher gas content exhibits larger reversal. Then, a mathematical model was developed to simulate the carbon isotopic reversals of methane and ethane. This model is based on these hypotheses: (i) diffusion flow is dominating during gas transport process; (ii) diffusion coefficients are nonlinear depending on concentration gradient. Our model not only shows a good agreement with isotopic reversals, but also well predicts gas production rates by selecting appropriate exponents m and m∗ of gas pressure gradient, where m is for 12C and m∗ is for 13C. Moreover, the (m−m∗) value has a positive correlation with fractionation level. (m1−m1∗) of methane are much higher than that of ethane. Finally, the predicted carbon isotopic reversal magnitude (δ13C1−δ13C2) exhibits a positive relationship with total gas content since gas in shale with higher gas content experiences a more extensive high Kn number diffusion flow. As a result, our model demonstrates an impressive agreement with the experimental carbon isotopic reversal data.