Environmental disasters result in the death of tens to hundreds of thousands of people and the loss of US$250-300 billion annually. Vulnerability to environmental disasters stems from both social and biophysical factors. While there is increasing awareness that individual hazards are often found in combination with other environmental or social risks in what can be referred to as multi-hazard landscapes, few studies directly examine how people respond to environmental hazards in a multi-hazard environment and the role that risk reduction and development organizations (DOs) play in that response. In this dissertation, I address this research gap through an investigation of risk perception and management in a multi-hazard environment of eastern Uganda dominated by people relying on subsistence agriculture for their livelihoods. Using a combination of qualitative and quantitative statistical analyses, I investigate how individual farmers and DOs differ in their perception and prioritization of hazards and the factors that influence farmers’ perception of multiple risks and their decisions to adopt best management strategies. Building on this household-level analysis of perception and action, I also draw on data from community-level focus groups and participatory mapping exercises to relate individual to community vulnerability. Results from these analyses show that the factors that shape farmers’ perception and management of different environmental hazards are not universal. Instead, the predictors of risk perception and adoption of best management practices are unique to particular hazards and management strategies. DOs can play an important role in reducing vulnerability through training and material inputs but need to recognize the heterogeneity of communities in doing so. Results show that communities are heterogeneous with respect to vulnerabilities, motivations, and capacities. DO programs must address these differences to achieve perception and behavior changes on a large scale. Participatory mapping exercises can be useful complements to expert risk assessments as they highlight local capacity and risk prioritizations, which do not always align with those determined by outside experts. While mapping is a promising tool for vulnerability analysis, the aspatial and unmappable components of vulnerability require a combination of methods across many scales and data types in order to be more holistically understood.