Global debates on climate change and tropical forests tend to focus on large-scale patterns of forest disturbance and carbon emissions. However, an equally important concern is to reduce the human impacts of climate change by improving the adaptive capacity of poor, marginalized populations to severe weather events such as floods or droughts. Hydro-climatic extremes probably have the greatest negative impacts in regions such as the Amazon, where floodplain agriculture is critical to regional food security and river transport is the main way of distributing food to remote road-less cities. For example, recent Amazonian floods and droughts have led to disease, disrupted food distribution, deaths from malnutrition and declarations of states of emergency. In this seminar, Luke overviews a UK-Brazil research initiative whose objective is to build the resilience of Amazonian cities to extreme hydro-climatic events. Initially, by developing tools to predict the likelihood of food insecurity and malnutrition in road-less Amazonian cities. Using statistical techniques developed by epidemiologists, they will develop real-time Food Insecurity Risk Maps, in order to assist strategic decision-making and resource investment to the impacts of extreme events. A range of data are being collected to inform the development of these predictive models. For instance, GIS and interviews are combined to show that riverine cities become, in effect, more remote during droughts. On a local-scale, Luke examines whether there are urban food deserts in vulnerable rainforest cities. Linking cities, he presents evidence that long journey times from regional centres lead to higher food prices, exposing the urban poor to food insecurity. He refers to a range of existing data on hydrological regimes and human health indicators; food availability and prices; and to-be-collected data on household-level deprivation, food security and coping mechanisms. Finally, he considers whether urban-rural linkages such as wildlife harvest and consumption might provide effective, sustainable forms of natural insurance for coping with extreme events.