Children comprise the most innocent and vulnerable group of people suffering from poverty, and the consequences of child poverty could be detrimental and permanent. Although the literature on child poverty in China is increasing, very little is known about the spatial disparities and influencing factors of child multidimensional poverty in China. To help fill the knowledge gap, this study used spatial autocorrelation analysis and investigated the geographical disparities of child multidimensional poverty in China from 2010 to 2016. Then, the trend of child poverty was predicted through Markov chain analysis. In addition, the factors affecting the regional and provincial differentiation of child multidimensional poverty were examined by the Geodetector method. Results show that (1) Child multidimensional poverty has demonstrated a downward trend since 2010, and the improvement of health has made a greater contribution to the reduction of child poverty than has that of living standards, care and protection and education. (2) Children in the eastern region have low levels of multidimensional poverty, while child poverty is relatively severe in the western region. Among the 25 provinces, child poverty in Sichuan has been the worst from 2012 to 2016. (3) Family conditions, economic development, urbanisation level, support ability and cultural environment are important factors affecting the spatial disparities of child multidimensional poverty. The findings of this study provide essential implications for child poverty reduction in China.