Identifying the driving forces that cause changes in forest ecosystem services related to water conservation is essential for the design of interventions that could enhance positive impacts as well as minimizing negative impacts. In this study, we propose an assessment concept framework model for indirect-direct-ecosystem service (IN-DI-ESS) driving forces within this context and method for index construction that considers the selection of a robust and parsimonious variable set. Factor analysis was integrated into two-stage data envelopment analysis (TS-DEA) to determine the driving forces and their effects on water conservation services in forest ecosystems at the provincial scale in China. The results showed the following. 1) Ten indicators with factor scores more than 0.8 were selected as the minimum data set. Four indicators comprising population density, per capita gross domestic product, irrigation efficiency, and per capita food consumption were the indirect driving factors, and six indicators comprising precipitation, farmland into forestry or pasture, forest cover, habitat area, water footprint, and wood extraction were the direct driving forces. 2) Spearman's rank correlation test was performed to compare the overall effectiveness in two periods:stage 1 and stage 2. The calculated coefficients were 0.245, 0.136, and 0.579, respectively, whereas the tabulated value was 0.562. This indicates that the driving forces obviously differed in terms of their contribution to the overall effectiveness and they caused changes in water conservation services in different stages. In terms of the variations in different driving force effects in the years 2000 and 2010, the overall, stage 1, and stage 2 variances were 0.020, 0.065, and 0.079 in 2000, respectively, and 0.018, 0.063, and 0.071 in 2010. This also indicates that heterogeneous driving force effects were obvious in the process during the same period. Identifying the driving forces that affect service changes and evaluating their efficiency have significant policy implications for the management of forest ecosystem services. Advanced effectiveness measures for weak regions could be improved in an appropriate manner. In this study, we showed that factor analysis coupled with TS-DEA based on the IN-DI-ESS framework can increase the parsimony of driving force indicators, as well as interpreting the interactions among indirect and direct driving forces with forest ecosystem water conservation services, and reducing the uncertainty related to the internal consistency during data selection.