The continental clastic reservoirs mainly distributed in the eastern region of China are regarded as the “ballast stone” for ensuring national energy security. After long-term waterflooding, reservoir petrophysical parameters have changed drastically. When the ultra-high water-cut period is achieved, the preferential flow paths are widely distributed inside the reservoir, resulting in severe ineffective water circulation and high difficulty in inhibiting water cut rise and stabilizing oil production. Due to the slow calculation speed and low identification accuracy of the existing methods, it is crucial to propose a novel method to identify preferential flow paths by considering the time-varying impact of petrophysical parameters in ultra-high water-cut reservoirs. To tackle this issue, massive dynamic and static data are collected from typical water-drive reservoirs in the Daqing oilfield to analyze the dynamic characteristics of preferential flow wells and the time-variation laws of reservoir parameters. A rapid evaluation index system for preferential flow paths is thereafter established. By describing the time-variation of reservoir permeability and fluid apparent viscosity and introducing a dynamic flow resistance coefficient to infer the inter-well or inter-layer connectivity, a novel method with the time-varying effect of petrophysical parameters considered is developed to quickly identify preferential flow paths in ultra-high water-cut reservoirs. A critical criterion of preferential flow paths is further constructed. Finally, the proposed method is used to determine the spatial distribution of preferential flow paths in a typical block of the Daqing oilfield. Results demonstrate that, once a preferential flow path is formed, some distinct dynamic characteristics of injectors and producers will be exhibited. Based on the estimated dynamic flow resistance coefficient, the preferential flow paths are further classified into four types: non-preferential, primary preferential, intermediate preferential, and strong preferential flow paths. Using the proposed method, the identification accuracy of preferential flow paths exceeds 95%, and the identification speed is more than 10 times faster than the traditional methods.