Abstract: In-depth understanding the variation of vegetation spring phenology is important and nece-ssary for estimation and prediction of ecosystem response to climate change. Satellite-based estimation is one of the important methods for detecting the vegetation spring phenology in Northern Hemisphere. However, there are still many uncertainties among different remote sensing models. In this study, we employed NDVI satellite product from 1982 to 2009 to estimate vegetation green-up onset dates in spring across Northern Hemisphere, and further analyzed the phenology spatio-temporal variation and the relationship with climate. Results showed that spatial mean spring phenology significantly advanced by (4.0±0.8) days during this period in the Northern Hemisphere, while spring phenology advanced much faster in Eurasia (0.22±0.04 d·a^-1) than in North America (0.03±0.02 d·a^-1). Moreover, phenology of different vegetation types changed inconstantly during the period. All five methods consistently indicated that grassland significantly advanced, while forests didn’t advance robustly among methods. In addition, the interannual change of spring phenology was mainly driven by spring temperature. The spring phenology advanced (3.2±0.5) days with 1 ℃ increase in temperature. On the contrary, we did not find significant relationship between vegetation spring phenology and spring accumulative precipitation across the Northern Hemisphere (P>0.05) in this study.