Taking the upper Lijiang River Basin, a critical area for flood control and water resources security in Guilin, as the study area, the daily measured rainfall data of 38 rainfall observation stations from 1983 to 2023 were converted into areal rainfall data of 13 spatial sub-basins based on the Thiessen polygon method. The Mann-Kendall trend test and the extreme-point symmetric mode decomposition (ESMD) method were employed to analyze the temporal and spatial evolution characteristics of rainfall across four temporal scales: the entire hydrological year, the main flood season, the post-flood season, and the dry season. The spatial and temporal unevenness of rainfall was revealed based on the temporal and spatial variation coefficients. The results indicate that the annual rainfall and the rainfall during the main flood season increased significantly at rates of 11.36 mm/a and 9.10 mm/a, respectively, while the rainfall in the post-flood season and the dry season showed no significant trend, which enhanced the risk of flooding during the flood season in the upper Lijiang River Basin. The annual distribution of rainfall in the basin was spatially and temporally uneven, and the average temporal variation coefficient of the 13 spatial sub-basins was 2.174, indicating a non-significant increasing trend in temporal inhomogeneity and a widening disparity between wet and dry periods. Spatial variation coefficients exhibited a slightly decreasing trend throughout the entire year, the main flood season, and the dry season, suggesting the spatial distribution of rainfall tends to be more uniform, while spatial inhomogeneity increased slightly during the post-flood season, complicating the regulation of reservoir groups during this period. The rainfall in the upper Lijiang River Basin has exhibited distinct and inconsistent changes at different temporal and spatial scales. These changes increase the complexity of flood control and beneficial operation for the upstream reservoir groups, elevate the challenge for urban flood control in Guilin, and introduce greater uncertainty in the ecological landscape flow guarantee of the Lijiang River.