Using changes in ground temperature to reflect the flow status of groundwater is one of the methods for predicting mine water inrush. In this study, in order to make this method suitable for different geological conditions, an improved method for predicting mine water inrush is established based on the theories of heat transfer and nonlinear water flow in fractal porous media. A water inrush judging criterion based on the critical pressure gradient of nonlinear flow is first established. Then, an internal structural model of the crushed rocks and a mathematical model of nonlinear flow in crushed rocks are derived based on the fractal theory. Finally, a thermal, hydraulic, and mechanical (THM) coupling model is established to study the nonlinear water inrush process and temperature changes. The improved method is established based on the numerical simulation results of the THM coupling model. Results show that the water inrush judging criterion can simultaneously consider the water‐resisting capacity of intact and crushed rocks and quantitatively calculate the water‐resisting capacity of crushed rocks compared with the traditional method. The improved method is suitable for different cases with different water‐resisting capacities, ground temperature change ranges and gradients, and aquifer water pressures, which can improve the applicability of using ground temperature to predict mine water inrush.