To investigate the differences in drought propagation under various underlying surface conditions in Southwest China, the K-prototypes clustering method was employed to classify the study area into three underlying surface types: the mild karst–agriculture-dominated ecological area, the typical karst–forest and shrub ecological area, and the mild karst–alpine grassland and shrub ecological area. Drought propagation times were quantified using multiple drought indices in combination with the maximum correlation coefficient method, and structural equation modeling (SEM) was applied to construct drought propagation chains for different underlying surface types. The results showed that the standardized precipitation evapotranspiration index (SPEI) was strongly positively correlated with the standardized soil moisture index (SSMI) and the standardized runoff index (SRI) in Southwest China, while all drought indices showed moderate correlations with the GRACE-based groundwater drought index (GGDI), with maximum correlation coefficients ranging from 0.3 to 0.6. The propagation time from various drought types to groundwater drought in the typical karst–forest and shrub ecological area was shorter than that in other regions, with mean propagation times ranging from 6.9 to 7.8 months. Drought in Southwest China generally followed meteorological – shallow soil moisture – hydrological – deep soil moisture – groundwater propagation sequence. However, only in the typical karst–forest and shrub ecological area can deep soil moisture drought significantly propagate to groundwater drought, while in other regions the propagation trend toward groundwater drought was not significant.