Abstract: Hazardous weather, such as thunderstorms, icing, and turbulence, is one of the main causes of flight accidents. To avoid possible risks caused by temporary diversion of hazardous weather during flight, numerical and probabilistic forecasting are employed to predict en route hazardous weather at the flight planning level. Considering the uncertainty of hazardous weather, a flight path planning method for uncertain weather is developed that can guarantee flight operation safety to the greatest extent. First, using probabilistic meteorological forecast data, we establish a mapping relationship between the diagnostic elements of thunderstorms and the probability of their occurrence. This is realized by an ingredients-based methodology that considers the three occurrence conditions of thunderstorms: water vapor, instability, and lifting trigger conditions. Afterward, the C-F model is applied to integrate the probability data of thunderstorm diagnostic elements. This fusion process enables the calculation of thunderstorm probabilities and the establishment of a comprehensive thunderstorm area map. The icing prediction index and turbulence prediction index are then measured based on numerical prediction data, and the icing and turbulence regions are defined. To solve the problem that current studies only consider a single hazardous weather event, a rasterized risk map is crafted to determine the risk of hazardous weather by integrating regions prone to thunderstorms, icing, and turbulence. Based on this, traditional path planning algorithms are enhanced to maximize flying safety. For example, we propose a risk minimization Dijkstra algorithm and a risk minimization A* algorithm considering the risk of nonobstacle environments. Finally, using the forecast data of severe convective weather in Central China on April 3, 2023, a real risk map is created. Using this map, both enhanced risk minimization algorithms and traditional A*, Dijkstra, and RRT(Rapid-exploration random tree) algorithms are applied independently for path planning. The risk and length of each planned route are then calculated to evaluate the algorithm’s performance. The results show that the risk minimization A* algorithm minimizes the total risk of the path, the traditional algorithm minimizes the path length, and the risk minimization Dijkstra algorithm exhibits the best overall performance. Therefore, if the purpose of path planning is to minimize flight risk, the flight path based on the risk minimization A* algorithm should be selected because of its superior safety. When considering both path length and risk comprehensively, the flight path under the risk minimization Dijkstra algorithm is preferred because this approach proves to be more economical than the risk minimization A* algorithm. The RRT algorithm is more suitable for path planning in high-dimensional complex environments and has no evident advantage in solving the problems described in this work.