Terrestrial vegetation ecosystem plays an important role in maintaining the sustainable development of global ecosystem, providing strong supports in exchanging energy and substances between pedosphere and atmosphere. With the rise of global change researches, vegetation research becomes an important content of land use and cover change research, which is one of the branches of studies on global changes. And it is a key step to understand the complex relationships between the vegetation distribution and its physical and artificial processes by using synthetical, quantitative and multi-scale means. In this research, Normalized Differentiation Vegetation Index (NDVI) was calculated as an indicator of vegetation, and three environmental variables were selected as the influencing factors including two topographic variables (elevation and slope) and one human variable (land use intensity, LUI). Continuous wavelet based methods were used to investigate the hierarchical structures and scale-location correlations between NDVI and other environmental factors. And two transects were selected along the east- west orientation (transect A) and the north- south orientation (transect B) in subtropical mountainous and hilly region, Zhejiang, China. The results showed that: (1) Four scales about the spatial distribution of NDVI and its driving factors were identified from the wavelet variance curves, with stronger variance values at larger scales; (2) The characteristic scales along the east-west transect were around 40 km and 80 km, whereas the characteristic scales along the north- south transect were around 30 km and 50 km. In these scales, the distributions of vegetation cover, landform and human activities exhibited richer structure information; (3) Topographic and humanity factors affected the distribution of NDVI at different scales, which was concluded from the wavelet coherence figures. It was proved that the humanity was the major influencing factor in larger scales (>8 km), while the terrain factors had more important effects in relative smaller scales (0- 8 km). Generally, wavelet based approach provides an effective tool to analyze the multi-scale structure of ecosystem and to understand the scale-location specific relationships between environmental factors.