In this study, the sensitivities of net primary production (NPP), soil carbon, and vegetation carbon to precipitation and temperature variability over China are discussed using the state-of-the-art Lund-Potsdam-Jena dynamic global vegetation model (LPJ DGVM). The impacts of the sensitivities to precipitation variability and temperature variability on NPP, soil carbon, and vegetation carbon are discussed. It is shown that increasing precipitation variability, representing the frequency of extreme precipitation events, leads to losses in NPP, soil carbon, and vegetation carbon over most of China, especially in North and Northeast China where the dominant plant functional types (i.e., those with the largest simulated areal cover) are grass and boreal needle-leaved forest. The responses of NPP, soil carbon, and vegetation carbon to decreasing precipitation variability are opposite to the responses to increasing precipitation variability. The variations in NPP, soil carbon, and vegetation carbon in response to increasing and decreasing precipitation variability show a nonlinear asymmetry. Increasing precipitation variability results in notable interannual variation of NPP. The sensitivities of NPP, soil carbon, and vegetation carbon to temperature variability, whether negative or positive, meaning frequent hot and cold days, are slight. The present study suggests, based on the LPJ model, that precipitation variability has a more severe impact than temperature variability on NPP, soil carbon, and vegetation carbon.