Drought strongly reduces vegetation productivity and is expected to worsen with climate change; however, long-term patterns and mechanisms of the drought sensitivity of vegetation (S-dro) remain unclear for arid-semiarid Northwest China (NWC). Using satellite GPP and soil-moisture records (1982-2018), in this study, we 1) quantified spatiotemporal drought trends, 2) assessed S-dro across major vegetation types, and 3) attributed S-dro changes using a hierarchical segmentation mixed-effects model. We found a small but significant regional increase in drought (0.5 & times; 10(-3) yr(-1)), concentrated in the southeastern Tibetan Plateau and the Qilian and Tianshan Mountains, while declines were found in the Loess Plateau, Guanzhong Plain, and Ordos Basin. Overall, 52% of NWC shows positive drought sensitivity, and S-dro increased regionally at 0.004 yr(-1) (p < 0.05); 48% of the area experienced significant rises in sensitivity. Sensitivity patterns were found to differ by vegetation: alpine meadow and alpine grassland exhibit the highest proportions of positive S-dro (73.2% and 63.3%), but only alpine meadow showed a declining trend; moreover, croplands showed the largest increase in S-dro. Vapor pressure deficit (VPD) and precipitation (Pre) are the dominant controls (VPD was positively correlated with S-dro in 89.1% of the area; Pre was negatively correlated with S-dro in 81.1% of the area), with alpine meadows mainly influenced by Pre and temperature and temperate grasslands by solar radiation. These results clarify where and why vegetation in NWC is becoming more drought-sensitive, thus providing targeted information for regional drought management and ecosystem resilience planning.