Free search (FS), which was set up based on the group of animal behavior, was adopted as an optimization algorithm to calibrate the Sacramento (SAC) model and the Xin’anjiang (XAJ) model. Calculation of the calibrated SAC model and XAJ model were conducted for the three watersheds of Songhua River System and Liao River System. Applicability of the SAC model was validated in cold and arid regions of Northeast China via comparison of the simulation results between the SAC model and XAJ model. The results indicated that the iterative calculation process of FS was simple and needed a few settings of the algorithm parameters, and it exhibited relatively high efficiency in the process of SAC model and XAJ model calibration; the simulation result of the SAC model was better than that of the XAJ model under the same conditions and its Nash—Sutcliffe model efficiency coefficient was higher than that of XAJ model. Whereas, the simulation accuracy of the SAC model needed to be improved in the case of application to the cold and arid region of Northeast China, and the performance of the SAC model needed further development and improvement.