Most existing soil carbon cycle models have modeled the moisture and temperature dependence of soil respiration using deterministic response functions. However, empirical data suggest abundant variability in both of these dependencies. We here use the recently developed SUPECA (Synthesizing Unit and Equilibrium Chemistry Approximation) theory and a published dynamic energy budget based microbial model to investigate how soil carbon decomposition responds to changes in soil moisture and temperature under the influence of organo-mineral interactions. We found that both the temperature and moisture responses are hysteretic and cannot be represented by deterministic functions. We then evaluate how the multi-scale variability in temperature and moisture forcing affect soil carbon decomposition. Our results indicate that when the model is run in scenarios mimicking laboratory incubation experiments, the often-observed temperature and moisture response functions can be well reproduced. However, when such response functions are used for model extrapolation involving more transient variability in temperature and moisture forcing (as found in real ecosystems), the dynamic model that explicitly accounts for hysteresis in temperature and moisture dependency produces significantly different estimations of soil carbon decomposition, suggesting there are large biases in models that do not resolve such hysteresis. We call for more studies on organo-mineral interactions to improve modeling of such hysteresis.