Abstract: Based on the carbon emissions and social-economic data of 62 counties in Zhejiang Province from 2014 to 2020, the spatial dependence and agglomeration of county-level carbon emissions are analyzed through the spatial auto-correlation test and local spatial auto-correlation test respectively. According to the spatial-temporal characteristics of county-level carbon emissions revealed by the index of Moran′s I and Local Moran’s I, the spatial error STIRPAT model is used to study the influencing factors of county-level carbon emissions in Zhejiang Province, China. The main results are as follows: (1) The total amount of county-level carbon emissions in Zhejiang Province fluctuates and shows a growth trend. (2) County-level carbon emissions have a significant spatial correlation, and the spatial agglomeration trend is relatively stable, which is consistent with the hypothesis of the geographical polarization effect. (3) High-high agglomeration counties are concentrated in the northeast of Zhejiang Province, while low-low agglomeration counties are mainly in the southwest. (4) The relationship between county per capital GDP and carbon emissions has not been "decoupled"; the low-carbon effect brought by large-scale industrial development as well as scientific and technological innovation has not yet appeared. (5) The carbon emissions of a single county is positively affected by the carbon emissions of the neighboring counties, and other social-economic factors affecting carbon emissions among counties also have a spatial correlation. Therefore, the policy of realizing regional coordinated development and the carbon peaking and carbon neutrality goals should not only focus on industrial layout, but also pay attention to dynamic and comprehensive consideration from a spatial perspective.