This paper analyses the impact of aid, and its optimal allocation, when conditionality is ineffective. It is assumed that the recipient government will implement its own preferences no matter what. In this set up, aid can still affect the behavior of a recipient, not through conditionality but through changing resource constraints. We analyze the problem in the tradition of models of optimal non-linear income taxation. We find that unconditional aid increases national income and makes the poor better off in the recipient country, but that there is a crowding out effect as the recipient country reduces labor supply in response to increased aid. On optimal allocation of aid across countries, we find that poorer countries should get more aid, as should countries with governments that are more inequality averse, which conforms to intuition. However, a striking finding is that more unequal countries should get less aid.