Along with the improvement of social productivity and living standard, residential buildings generate a growing portion of carbon emissions, especially during the operation stage. However, energy use behaviors are usually ignored in carbon emission calculation. This study focuses on calculating carbon emissions during the operation stage for residential buildings based on the characteristics of energy use behaviors in different regions. Firstly, we investigated energy use behaviors in dwellings across three cities in China: Xi’an, Shanghai and Fuzhou. Then, we established calibrated carbon emission models and optimization models with different green building measures for residential buildings. The results of this research reveal a significant disparity between the energy usage habits of residents in different climate regions. The carbon emissions of residential electricity bills in Xi’an, Shanghai and Fuzhou are 13.6 kgCO₂/(m²·a) (excluding central heating), 29.3 kgCO₂/(m²·a) and 17.2 kgCO₂/(m²·a), respectively. Equipment carbon emissions account for 32.2%–64.1% of the total. In comparison to the model based on internal standard setting, the accuracy of the models using actual internal has improved by 25.9%–37.4%. The three-star green building methods have the highest carbon reduction rate among different star buildings, the emission reduction rates are around 30%. This study’s findings are useful for carbon emission calculation and green building design of residential buildings in the future.