Abstract: Based on CASA model, net primary productivity (NPP) of vegetation was estimated from 1982 to 2019 in Yunnan Province, and its spatiotemporal variation characteristics were analyzed. In combination with the standard precipitation evapotranspiration index (SPEI), the relationship between droughts and NPP at different time scales was discussed. The results show that as the SPEI accumulation increased in the study area, so did the proportion of areas with drought frequency greater than 30%. NPP is relatively low in the years with large-scale droughts. There was a negative correlation between average annual NPP and drought affected area, mainly concentrated in central Yunnan Plateau region. When the proportion of drought area increased from under 30% to 30%−40%, 40%−50% and above 50%, NPP decreased by 3.3%, 4.5% and 8.7% respectively. Long-term drought on a 12-month scale had a greater impact on NPP than short and medium term drought. On the whole, the sensitive period of NPP to drought appeared in May and October, but the sensitivity responses of different vegetation types to drought were different. Tropical forests were most sensitive to three-month drought in June and to 12-month drought in August. Meanwhile, shrub, grassland and agricultural vegetation were the most sensitive to 3-month drought, while alpine vegetation was the most sensitive to 6-month drought.