Analyzing the incidence of stand mortality, constructing a predictive mortality model, and identifying the factors influencing mortality can provide an evidence-based framework for managing forest growth and productivity. This study investigated the mortality of Chinese fir (Cunninghamia lanceolata (Lamb.) Hook.) plantations in Jiangxi Province, a subtropical region of China, utilizing national forest inventory data from 2006 to 2016. A total of 18 potential factors were identified across three categories: stand, site, and climate variables. Four statistical models were evaluated, including the zero-inflated Poisson model, zero-inflated negative binomial model, hurdle Poisson model, and hurdle negative binomial model, with incorporation for random effects at the plot-level to determine the most effective mortality model. The results showed that incorporating random effects at the plot-level significantly improved the fit and predictive capabilities, with the zero-inflated negative binomial model, which accounts for random effects in canopy cover, performing best. Mortality increased with higher canopy cover, stand basal area, elevation, and mean warmest month temperature, but decreased with greater humus thickness. It is advisable to maintain litter within the plantations to increase humus thickness. Moreover, implementing strategic thinning is vital for reducing inter-tree competitive pressures, managing microclimate conditions to promote temperature reduction and humidity enhancement during hot seasons and temperature elevation and humidity increase during cold seasons. These management approaches are projected to decrease mortality.