In order to understand why development projects fail on a regular basis - or at least fail to achieve their intended goals - it is important to investigate not only the apparatuses of development (work related aspects such as documents, reports, policies, programmes and projects), but also the social and cultural worlds of the different actors involved. This means expanding the analysis beyond critiques of development that focus on discourse analysis of project documents and interface analysis which produce in-depth ethnographies of those being "targeted' for development; it means also focusing attention on those who are doing the actual "targeting'. This thesis studies how different groups of actors engage in the same material reality - that of the natural environment in rural Madagascar. It is a multi-sited ethnography that explores the culture and lives of Betsileo farmers living and working in the Central Highlands of Madagascar, as well as various development practitioners living and working on development projects and programmes concerned with the management of those same natural resources and livelihoods. The research focuses on three inter-related sites and traces development relations and interactions from the central locations of the development machine to its extremeties: several villages around one rural town in the Central Highlands called Ambalavao, the provincial capital of Fianarantsoa, and the capital of the country, Antananarivo. And it maps out the social and work relations of various actors as they engage in rice cultivation and the management of dams and irrigation works with various degrees of knowledge, capacities and power. The focus is not on anyone development project or programme, but the various relationships - or lack thereof - between different actors, domains of science, knowledge and various meanings given to them. It is also a commentary on the wider global and bilateral political-economic relations that both limit and structure what development ends up doing - or not doing.