Why do the humanitarian principles, human rights and other aims espoused by NGOs apparently fail to influence the reality of assistance delivery, whilst reality does not dint these objectives. Drawing on literature concerning assistance, the political economy in which it is given, and psychological dimensions of regulation, I investigate the nature and function of objectives stated by aid providers. I argue that whereas discourse is manoeuvred to approve whatever NGOs undertake, the assistance delivered is so far from its objectives as to question what guidance they offer and their strategic merit. Additionally, the NGOs' approach disregards failure by allowing operational weaknesses to blame or discredit others, rather than being acknowledged to reflect on interventions. This screens assistance from evaluation or reform, whilst overlooking on-going suffering. My methodology combines an analysis of donor and NGO policy-making with an empirical investigation of how people are given, or excluded from, assistance. I find that aid agencies maximise returns on a politically constructed morality. Where these are high, as in the cases of Sierra Leone and Rwanda, a consensus of sorts is reached between aid providers and governments, and some people benefit from assistance. The consensus, though, marginalises opposition groups and people in their territory. In the Democratic Republic of Congo and southern Sudan, no consensus emerged and the contexts offered nothing to aid providers; most people received no genuine help. In all countries, NGOs deemed political and military activity illegitimate, thereby forestalling dialogue and limiting their perception of the situation; these factors cast doubt on the sincerity of the interventions. I conclude that people in countries at war are not 'breaking the rules' of assistance - as assistance is not meaningfully 'ruled' by rights or principles - they are more fundamentally 'not playing the game'.