Although the IMF presents itself as a monetary institution, it plays an important role in providing support to poor countries via its Poverty Reduction and Growth Facility. It is difficult to imagine more central development issues than poverty and growth. However, while there is a broad consensus surrounding the stabilization issues with which IMF programmes conventionally deal, there is much less agreement about the causes of economic growth and poverty. This carries lessons for the design of the PRGF. While most reviews of it focus on 'process', this paper offers a more fundamental analysis of the Fund's involvement in growth and poverty reduction. It suggests that the design of conditionality needs to distinguish between elements according to the degree of scientific consensus. It also suggests that the success of the Fund's attempts to facilitate growth and poverty reduction will be severely constrained unless the necessary external, financial support is provided. © 2004 John Wiley and Sons, Ltd