Global climate change has intensified the frequency and severity of drought, profoundly impacting vegetation phenology in Xinjiang arid zone. We employed Solar-Induced Fluorescence (SIF), MODIS phenology product and the Standardized Precipitation Evapotranspiration Index (SPEI) from 2001 to 2020 to extract the spatiotemporal evolution of vegetation Start of Season (SOS) and End of Season (EOS) and conducted a quantitative analysis of the cumulative and lagged effects of droughts on vegetation SOS and EOS, as well as the characteristics of drought resilience. The results indicated that: (1) Comparison with the phenological product revealed that vegetation SOSGOSIF was delayed by an average of 13.450 days compared to SOSEVI, while EOSGOSIF was advanced by 24.257 days compared to EOSEVI; (2) The overall trend of vegetation SOS in Xinjiang had shown a significant advancement, while the EOS had exhibited a delayed trend; (3) SOSGOSIF was predominantly suppressed by 1-month drought cumulative effects, whereas EOSGOSIF exhibited higher sensitivity to 6-month drought. Furthermore, SOSGOSIF showed a 2-month lagged response to drought and EOSGOSIF had a 1-month lagged effect; (4) The resilience of different vegetation phenologies to drought exhibited notable variation. Grassland SOSGOSIF demonstrated the strongest post-drought recovery capacity, whereas shrubland EOSGOSIF exhibited the highest drought resilience. This study highlights the advantages of GOSIF in monitoring vegetation phenology in Xinjiang. It provides novel insights into the cumulative and lagged effects of multi-scale droughts on vegetation phenology in arid environments. Furthermore, we developed an integrated framework encompassing 'drought stress-phenological response-vegetation growth recovery' and quantified differences in the post-drought recovery capacity of both SOS and EOS across various vegetation types. These findings offer a scientific basis for ecological restoration initiatives and sustainable development in arid zones.