Should a donor delegate the responsibility for allocating its aid budget to a less inequality-averse agent to alleviate the consequences of the Samaritan's Dilemma it is facing? I show that when aid impact differs across recipients the optimal type of agent depends on whether or not committing to a greater share for countries where the productivity of aid is low raises the combined domestic incomes of recipients. This is the case for donors too concerned with efficiency ex post. They therefore delegate the decision on the discretionary aid allocation rule to agents more sensitive to distributional issues than themselves. (c) 2005 Elsevier B.V. All rights reserved.