The super-efficiency slacks-based measurement model was used to evaluate provincial water resources utilization performance in China during 2005-2020. The Dagum Gini coefficient decomposition method, spatial autocorrelation, and temporal and spatial transition analysis were used to analyze the spatial differentiation. The driving mechanism of spatial differentiation was analyzed by using geographical detector. The results showed that the level of provincial water resources utilization performance was generally not high. The degree of spatial differentiation had a trend of fluctuation and expansion on the whole, and the main source was supervariable density, followed by inter-regional differentiation and intra-regional differentiation. The spatial differentiation pattern showed a spatial correlation characteristic of “high in east and low in west”, with a significant spatial spillover effect and relatively stable local spatial distribution structure. Population size, economic level, and urbanization level had significant driving effects on the spatial differentiation of water resources utilization performance, and the effects were different in different regions. Water resources utilization performance was economically driven on the national scale and in the eastern region, and it was urbanization-driven and population-driven in the central region and western region, respectively. The driving effect of industrial development on spatial differentiation of water resources utilization performance had regional heterogeneity. Water resources utilization performance was industry-driven on the national scale and in the eastern region, while it was construction industry-driven and agriculture-driven in the central region and western region, respectively. It is suggested to formulate coordinated improvement strategies for water resources utilization performance in relevant regions according to the leading driving factors.