Drought critically affects ecosystem productivity and regional sustainability. Mountainous regions, characterized by complex topography and diverse vegetation types, are highly sensitive to climate warming and therefore serve as natural laboratories for investigating drought evolution and vegetation responses. Here, we systematically analyzed the spatiotemporal characteristics of multidimensional droughts (SPEI, SSI, and SRI) across the Tianshan Mountains. Based on the long-term NDVI series (1982-2022), vegetation-type-specific partial least squares path modeling was employed to elucidate the pathways through which drought and environmental factors influence vegetation dynamics. We found that: (1) The meteorological, soil and hydrological drought increased significantly during 1982-2022, with a significant increase around 2005. The areas affected by extreme meteorological, soil, and hydrological drought expanded by approximately 23%, 20%, and 16%, respectively. (2) Drought trends varied markedly among vegetation types, with meteorological drought intensifying across all vegetation types. Forests showed partial mitigation of soil and hydrological droughts, whereas agricultural land experienced the most severe hydrological droughts. (3) NDVI was primarily influenced by soil and hydrological droughts, with agricultural land and sparse vegetation showing the highest drought sensitivity, whereas forests exhibited the weakest response. (4) Spatial NDVI trends in forests showed limited sensitivity to drought, and those in sparse vegetation were associated only with SPEI. In contrast, NDVI trends in grasslands and agricultural land were positively associated with both SPEI and SRI trends. This study evidences the intensification of multiple droughts and reveals vegetation-specific drought responses in the arid and semi-arid mountainous regions, highlighting the need for ecosystem-specific climate risk assessments.