It is a paradox of communication and media studies that, while media are widely seen as key institutions in the “project of time, space and life management” (Silverstone, 2005), not enough attention is given to the ways in which mediation is socially produced and becomes politically effective. In the case of development communication intervention, although mediation is rarely taken into account as an analytical variable in studies, it is implicit in the practice (Enghel, 2014). Processes of mediation connect funders – via the governmental or intergovernmental agencies acting on behalf of the world’s “advanced countries”, in OECD jargon – to presumed beneficiaries – the citizens of “developing countries” – by way of the implementation of specific projects. In the act of framing a problem and favoring a solution, and at the same time communicating about problem and solution in particular ways, development communication intervention mediates the complex relationship between international funders, national governance structures (including, but not limited to, their media systems), and citizens on both ends, in the “giving” and “receiving” nation states. How do citizens on the “receiving” and giving” ends of international development communication encounter and experience media? What roles do the media play in interventions aimed at doing good? What roles do they play in communicating do-gooding? Is there a relationship between the forms of media engagement proposed at each end? Which media technologies are foregrounded in institutionally-driven international development communication, and which media practices are favored? Which models of engagement influence the choices of funders and implementers, and which kinds of recipients are implied in those choices? Through a qualitative study of the Videoletters project, this paper examines what ‘media engagement’ means for the specific case of international development communication (and by extension, for foreign affairs) and analyzes the theoretical and methodological challenges posed by its research. Videoletters was a high-profile media-driven intervention aimed at reconnecting ordinary people affected by ethno-political divisions across the former Yugoslavia, implemented between 2000 and 2005. Adopted by European bilateral funders for large-scale implementation, the project was categorized as a “tool for reconciliation”. The paper describes and analyzes a number of intricate ways in which ‘media engagement’ was understood, put into practice, received and evaluated in practice throughout the intervention.