Abstract: Steel spiral cases in pumped storage power stations are critical pressure-bearing structures. They are subjected to long-term multi-field coupling effects such as cyclic internal water pressure, thermal loads, and concrete constraints, making them susceptible to low-cycle fatigue damage. Traditional methods for compiling load spectra often consider only constant internal water pressure cycles while neglecting the influence of thermal stress, which may lead to overestimation of fatigue life. This study takes the spiral case of a specific pumped storage power station as an example and proposes a hybrid model-based method for compiling fatigue load spectra. By integrating finite element numerical simulation and statistical modeling, the contributions of reservoir water level and water temperature changes to stress are separated. The proportion of thermal stress is quantified and superimposed into the load spectrum. Three types of load spectra were compiled: the prototype spectrum, the rainflow-reorganized spectrum, and the constant-amplitude spectrum. Fatigue analysis software nCode DesignLife was used to predict the low-cycle fatigue life of the spiral case. The hybrid model demonstrated good fitting performance, indicating the rationality of the proposed method. After considering thermal effects, the predicted fatigue life decreased by approximately 60%, highlighting the necessity of incorporating temperature influence in fatigue load inputs. This study provides a reasonable load spectrum compilation method for the anti-fatigue design of steel spiral cases and offers a basis for predicting their low-cycle fatigue life.