Applying the data of industrial energy consumption from 1997 to 2011, the data of energy consumption per unit of GDP in districts or counties of Xi'an City in 2010 were estimated, and the model to analyze the carbon emissions driving factors was built by using STIRPAT model. The results show that all kinds of carbon emissions presented generally, an increasing trend during 1997-2011. Carbon emissions, carbon emissions output and per capita carbon emissions and carbon emissions per unit area of land presented significant regional differences in each district or county. Moran index showed that the spatial correlation of carbon emissions in each district or county was positive correlation, and the spatial distribution feature of carbon emissions was that:the district or county with higher carbon emissions tended to aggregate to the one's of the higher, the district or county of lower carbon emissions tended to aggregate to the one's of the lower. The STIRPAT model showed that:the influence degree of economic development, population scale, industrial structure and technical level on carbon emissions was different. Moreover, economic development played the decisive role on increasing carbon emissions, but the industrial structure optimization had inhibitory effect on the increasing of carbon emissions.