Abstract: Analysis of carbon emission based on small scale is necessary for making different policy for different city. Socioeconomic data and energy consumption data of each city in ZheJiang province are used to calculate the total carbon emission, carbon emission per capita and carbon emission per unit land area from 2005 to 2010. Spatial autocorrelation analysis is used to reveal the spatial correlation of carbon emission index between neighbor cities. Geographically weighted regression model is applied to describe the spatial heterogeneity of the influencing factor of total carbon emission among the cities. Results showed that: (1) Total carbon emission of Zhejiang province increased from 304.86 million tons to 495.59 million tons during 2005-2010. Carbon emission per capita and carbon emission per unit land area both had rising trend and obvious spatial difference. (2) Each carbon emission index (total carbon emission, carbon emission per capita and carbon emission per unit land area) was highly spatial correlated. Highhigh agglomeration appeared at the north and the east areas, and Lowlow agglomeration appeared at the south and the west areas. (3) Population and investment were important driving forces of total carbon emission and both showed spatial heterogeneity in Zhejiang province. Investment played an increasingly important role in the growth of carbon emission in the south and the west areas.