A machine learning method called the random forest approach was used to explore the relationship between forest productivity and stand and climate factors. Data was sampled from the long-term spacing trails of Chinese fir plantations in southern China. Results showed that the productivity of Chinese fir plantations increased with increasing value of the Gini coefficient and dominant height (Hd), while it decreased with increasing age (A) and stand basal area (BA). Furthermore, forest productivity was positively correlated with annual precipitation (AP) and summer mean maximum temperature (SMMT); in contrast, it was negatively associated with winter mean minimum temperature (WMMT) and annual heat-moisture index (AHM). Age had the greatest influence on forest productivity compared to a more secondary influence of climate factors. We found that older forests were more vulnerable to climatic stress and the productivity of forests with middle- and high- levels of competition behaved similarly, and was lower than forests with low level of competition intensity. Higher SMMT, AP and lower AHM would enlarge the gap of forest productivity under different levels of stand structure, competition intensity and site quality. Changes in site conditions had little effect on productivity when AP was lower than 1250 mm. Our findings will provide a good framework for Chinese fir plantation management under future climate change.