The ability for communities to "bounce back" from major disasters is essential for poverty alleviation and economic development. Termed "disaster resilience", this process is of particular importance in China as rapid economic expansion and urbanization has increased Chinese susceptibility to a number of major disasters, including the 2008 Wenchuan Earthquake. Earthquake-induced landslides represent a particular challenge to resilience as increased rates of landslide hazard may persist for many decades. The proposed research seeks to understand what controls this persistent landslide hazard and the processes that cause landslides to jeopardise recovery. To understand the recovery process and how it affects resilience, we will investigate the role of "social vulnerability" in modifying the response to earthquakes and their related hazards. We will assess the underlying drivers of social vulnerability and the spatio-temporal differences across Sichuan province. We will combine our estimates of landslide hazard and social vulnerability across the decade after the Wenchuan Earthquake, investigating both the spatial patterns of risk and how these change with time. To achieve these goals, we will focus our work on the areas affected by the Wenchuan Earthquake, where the Chengdu Institute of Technology-State Key Laboratory of Geohazard Prevention and Geoenvironment Protection has created an incredibly large dataset of landslide hazards since the earthquake. In collaboration with landslide scientists and social scientists at Cardiff University's Sustainable Places Research Institute, we will expand this dataset in two ways; (1) increasing the resolution of landslide hazard mapping to understand the relative role of aftershocks and rainfall in controlling hazard, and (2) using local census data to understand social vulnerability and how the interaction between social vulnerability and landslide hazards has changed in space and through time. The unprecedented detail of our data will enable us to develop a new probabilistic landslide hazard model that incorporates landslides caused by both aftershocks and rainfall events that can be applied across earthquake-prone China and perhaps even globally. Field data collected as part of this effort will help to constrain threshold values and so help support the construction of a landslide early warning system for Sichan. Finally, we will model the resilience of the built environment and key infrastructure through state of the art machine learning algorithms. As evidence of our commitment to improve the welfare of earthquake-prone China through better planning for disasters we will engage with an extensive network of governmental and non-governmental institutions. From the first day of the grant we will engage with organisations with interests in both science and policy to achieve this goal. We will also model resilience under different demographic and policy scenarios, using this as a tool to understand and communicate the challenges of building resilient communities.