There exist large differences in household CO2 emissions (HCEs) between income groups in China, and the magnitude of difference varies significantly by consumption categories and provinces. Current studies have estimated China's interregional carbon inequality, however, details of carbon inequality between income groups and across consumption categories as well as the future view are still limited. This paper uses the environmentally extended multi-regional input-output (MRIO) model linking with the household consumption survey data to calculate the HCEs, that is, the emissions embodied in goods and services across urban and rural income groups of China's 30 provinces in 2017. Carbon inequalities of HCEs are measured by Gini coefficient and the related drivers are obtained by decomposition analysis. For assessing the carbon inequality in 2030 and 2050, this paper analyzes the income elasticity of demand for various consumption categories. The results indicated that in 2017, the top 10% of the population induced 24.7% of the HCEs, while the bottom 46% of the population were responsible for 24.6% emissions. The overall carbon inequalities of China (i.e., 0.32 in 2017 and 0.41 in 2050), urban China (i.e., 0.31 in 2017 and 0.35 in 2050) and rural China (i.e., 0.27 in 2017 and 0.34 in 2050) increase until 2050 with the demand-side growth, which can be attributed to the effect of household expenditure. More in depth, the income elasticities of demand for carbon-intensive goods, such as housing, others and transport, are larger in affluent groups, which will result in higher expenditure by rich people on carbon-intensive goods with income increase so as to widening gaps of HCEs between poor and rich individuals.