Cities create good environments for the generation and diffusion of innovation because they can gather people with different talents, education, culture and language. Based on the panel data of 272 prefecture level cities from 2009 to 2012, spatial autocorrelation of cities’ innovation is analyzed by Moran’s I index. Then cities’ innovation spillover effect is measured through spatial weight matrix based on exponential decay and Spatial Durbin Model(SDM). And the direct effect and indirect effect of each independent variable on the dependent are decomposed by partial derivative method. The relationship of spillovers effect and geographic distance in China and three regions are explored. Results show that innovations in cities have positively significant spillovers effect and it is robust for other two spatial weight matrixes and different regions. The expenditure for R＆D, scientific research personnel, economic development and foreign direct investment have positive effect on innovation. Total effect, direct effect and indirect effect of the expenditure for R＆D and per capita GDP are obvious. Spillovers effect of total cities is positive for the geographic distance from 150km to 1000km. And it reaches the maximum at 650km. Spillovers effect of east, middle and west region reach the max value at 150km, 550km and 750km.