Aid quality is often investigated using cross-country statistical methods. The article takes a more institutional approach. It reviews British bilateral aid, in order to investigate the complex interaction between policy and practice in aid for poverty reduction. There has been a stated desire to increase the poverty focus of the programme. However, it is difficult to trace the effect of a new policy in the statistics. More important has been the influence of an external factor, the increase in the demand for emergency aid. This has risen from 2 per cent to 14 per cent in a decade, increasing the poverty focus of the programme, but for the ‘wrong’ reason. At the same time, the share of technical cooperation has increased sharply: it is hard to trace the poverty?reducing impact of this form of aid. Statistical analysis which ignores policy shifts and changes in aid composition may be misleading.