Recrystallization of ice crystals is a physical cause of deterioration in frozen foods during both storage and distribution, and therefore needs to be controlled and predicted for optimal storage and distribution of frozen foods. However, a systematic understanding of recrystallization is lacking and the accumulation of sufficient recrystallization experiments remains difficult due to the time-consuming storage experiments necessary to investigate recrystallization behavior. In response to this challenge, we showed that the concept of water mobility in a freeze-concentrated matrix is useful for predicting the recrystallization rate of ice crystals in model frozen foods. Thus, knowing the water diffusion coefficient, an index of water mobility, in a freeze-concentrated matrix would enable us to predict the recrystallization rate of ice crystals and limit the need for time-consuming storage experiments.