Abstract: The urban-rural gradient is an effective technique to examine spatial pattern and processes of urbanization. Quantitative studies on interaction between landscape pattern and function of urban ecosystem have become the core issues of landscape ecosystem in city. Based on remote sensing images, geographic information system (ArcGIS 10.2) and landscape pattern analysis (Fragstats 4.2), four indicators commonly used including physical variables and landscape metrics were calculated in this paper. Principal components analysis and multiple linear regression analysis were used to build the urban-rural gradient model to assess the urbanization level of central Kunming in a direction and the differentiation of landscapes. The results showed as follows: (1) Urbanization indicators can be used to describe the level of urbanization. The gradient of urban-rural showed spatial differentiation. (2) With the change of urban-rural gradient, the index values of urbanization showed a decreasing trend. The proportion of impervious underlying surface area (80%), density of road network (14 km/km2), landscape shape index and Shannon diversity index were the highest in the urban region. (3)The average of urbanization index UI in district of Kunming was showed that: Guandu＞Wuhua＞Xishan＞Panlong, and the variation coefficient was smallest for Guandu ,and largest for Wuhua. (4) The development of urban area in central Kunming is more concentrated, the dividing line between the suburbs and the city is not obvious. The main reason may be restricted by orographic factors.