The method presented is helpful for assessing effects of land use policy and unfolding the action mechanism and implementation effect of policies on land use changes. The backcasting method is based on the re-simulation of the historical real changes of land use. By observing the error between the real and simulated states, adjust the parameters for many times, and a variety of land use scenarios are obtained, and the scenario with the minimum and acceptable error is selected to assess the effects of land use policy represented by its parameters. The method is structured in four sequential stages, including expressing the policy action mechanism as the parameter combinations to control land use changes, and then translating the parameter combination that can reflect the effect of land use policies into the policy effectiveness. This process is realized by ArcGIS and GeoSOS-FLUS software developed based on Cellular Automata (CA) model. The method was successfully tested in the peri-urban region near Shanghai metropolitan area. The raw material is Land-Use/Cover (LULC) data of study area in 2000 and 2015. This method assessed the effects of land use policies during fifteen years, and analyzed the mechanism of effective policies, as well as the types and reasons of failure policies. This presented method is useful in land governance and the formulation and implementation of land use policy. •This article developed a backcasting approach to evaluate the effectiveness of land use policies on land use changes in a certain period of time.•The policy action mechanism was transformed into controllable, visible, and adjustable parameter combinations.•The method built an analytical framework for the assessment of land use policy effectiveness in any regions.