We used a structural equation model to analyze the relationship between environmental and forest stand factors and net primary productivity in Cunninghamia lanceolata forests. We collected 644 data points from 155 published studies on net primary productivity (NPP) measurements of Cunninghamia lanceolata forests. The environmental factors included mean annual precipitation (MAP) and mean annual temperature (MAT). The stand factors included age and density of trees. The correlations between NPP and environmental and stand factors were different. NPP was significantly positively correlated with both MAP and MAT, with correlation coefficients of 0.630 and 0.378 respectively. Conversely, NPP was significantly negatively correlated with both age and density, with correlation coefficients of -0.332 and -0.408 respectively. Each variable fitted a normal distribution after natural logarithmic transformation. We used a structural equation model to explore the relationship between NPP and MAP, MAT, age, and density. The results showed that the structural equation model was an excellent method to explain the relationship between environmental and stand factors, and NPP. MAP, MAT, age and density, all had an effect on NPP, with total path coefficients of 0.398 (P < 0.01), 0.746 (P < 0.01), -0.321 (P < 0.01) and -0.738 (P < 0.01), respectively. MAT and age had both direct and indirect effects on NPP, as MAT had a direct effect on MAP, and age had a direct effect on density. MAT and age directly affected NPP as well, and were therefore included as direct and indirect path coefficients in the structural equation model. The direct path coefficients of MAT and age were 0.494 (P < 0.01) and -0.700 (P < 0.01) respectively. The indirect path coefficients of MAT and age were 0.252 (P < 0.05) and 0.379 (P < 0.05) respectively. The structural equation model analysis indicated that MAP and MAT were the strongest positive drivers of NPP, whereas age and density were the strongest negative drivers of NPP. The structural equation model analysis also indicated that MAP, MAT, age, and density explained 62% of the variation in NPP of Cunninghamia lanceolata forests. We conclude that the structural equation model is the most appropriate approach to understand and predict ecosystem functioning, as understanding NPP requires an accurate assessment of large-scale patterns in NPP distribution and partitioning in relation to environmental and stand factors.