We build and implement a normative procedure to allocate international aid based on equality of opportunity concerning the risk of poverty. This is an alternative to Collier and Dollar's proposal (2001) which stresses the impact of aid on worldwide poverty reduction. The big problem with their approach, as regards distributive justice, is that it leaves very great inequality in poverty risk between inhabitants of countries with widely varying structural disadvantages. We draw on post-welfarist theories of social justice, especially those of John Roemer. However our proposal is very different to that of Llavador and Roemer (2001), which has serious methodological errors and reaches contradictory conclusions. Our proposed allocations, like those of Collier and Dollar, differ from current aid allocation by giving more to the poorest countries. Apart from this agreement, our equality of opportunity principle takes account of structural disadvantages to growth rather than quality of past policies. Our kind of allocation shares out poverty risks much more fairly among the world's population, while reducing global poverty almost as effectively as Collier and Dollar's.