The increase of carbon emissions has a direct impact on extreme weather disasters and global warming. Therefore, controlling and reducing carbon emissions is one of the important ways to promote ecological balance and regional sustainable development. The difference in urban land use patterns and the different speeds and levels of urban economic development will, of no doubt, lead to the differences in carbon emissions. The amount of carbon emissions from different land use structures and the levels of its efficiency will directly affect the sustainable development of the urban economy, which has aroused the attention of academics. This paper estimated the quantities of carbon emissions of different land types by using the land use structure data from 11 districts in Nanjing from 2005 to 2014, and further introduced the estimated carbon emissions into the DEA model as an undesirable output variable to estimate, compare and analyze the total factor carbon emissions productivity growth of land use structure in Nanjing by using the Hicks-Moorsteen index method. The major results were shown as follows. First, the main sources of carbon emissions for land use structure in Nanjing were the indirect carbon emissions from the construction lands. During the study period, the carbon emissions of land use structure showed a continuous growth trend and certain degree of spatial heterogeneity in the districts of Nanjing city. The severe emission areas were mainly concentrated in the central regions and there was a tendency to shift to the northern regions. Second, the total factor carbon emission productivity growth and its decomposition factors were lower than the traditional total factor productivity (TFP) growth, which means that the traditional TFP growth without considering the constraints of carbon emissions overestimated the actual productivity growth level of land use structure. The total factor carbon emission productivity growth showed a convergence trend with the characteristic of equilibrium development between regions during 2006-2014, due to the fact that the development concept of "low carbon and high efficiency" were deeply rooted in people's mind and the energy consumption per unit of GDP was continuously decreasing. The calculation results of model 2 showed that the total factor carbon emission productivity growth level of Jiangning district was the highest and that of Yuhuatai district was the lowest. Third, the decomposition results of total factor carbon emission productivity growth indicated that the technical efficiencies of all the districts in Nanjing were relatively low, the "technology catch- up" effects were not obvious, and the scale economy effects of land use began to appear, but the scope economy effects were not very significant. Fourth, total factor carbon emission productivity growth presented a positive spatial correlation and the characteristics of spatial agglomeration in all the districts of Nanjing. The districts of high and high (HH) agglomeration were increasing and the districts of low and low (LL) agglomeration were decreasing. Finally, some suggestions were put forward.