Abstract <p>Climate change and human activities threaten the ecological stability of natural rivers, and an in-depth analysis of changes in basin hydrological regime and their driving mechanisms is essential for water resource management. In this study, the nonlinear relationship between runoff and meteorological variables was investigated by using the cross-wavelet transform method, and the hydrological regime changes were quantitatively assessed by combining the index of hydrological alteration (IHA) with the flow duration curve evenness indicator (FDCev). A particle swarm optimization-long short-term memory (PSO-LSTM) model is used to quantify the differences in the impacts of different drivers at different time scales. The results show that the climate-runoff system in the Han River Basin had three important resonance cycles from 1975 to 2020, and the hydrological regime experienced a moderate degree of change in all groups of indicators except for group 3. The driving force contributions showed significant differences in different time scales: on the annual scale, climate factors dominated the runoff changes (78.25%); on the seasonal scale, human activities had a more significant influence in winter (62.98%); and on the monthly scale, the months of January, February, March, and December were more significantly influenced by human activities, and climate factors mainly drove the rest of the months. This study can provide a scientific basis for water resource management in the basin.</p>