This paper extends the theory of learning-by-exporting effect to the field of green productivity, and studies the nonlinear effects of learning-by-exporting on enterprises’ green productivity. This paper explores the threshold effect of learning-by-exporting on green productivity through panel quantile regression and panel threshold regression model by using the data of China’s manufacturing enterprises. The results show that the effects of learning-by-exporting on green productivity are positive, and there are three thresholds effects of learning-by-exporting on green productivity. Learning-by-Exporting effect on green productivity show a typical Kuznets inverted "U-shaped curve which increases first and decreases then during both benchmark and robustness tests. This reflects the enterprises can improve their green productivity in a nonlinear mode through the learning-by-exporting effect.