This report sets out some broad ideas about how poverty evaluation could be conducted for ACIAR research projects. As with good benefit–cost analysis, there are good practices that need to be observed when undertaking poverty analysis. While poverty is a broad concept, and can be addressed through many means, these need to be grounded in some common understanding of the economics of poverty. This report is concerned mostly with quantitative evaluation, in the same sense that current ACIAR project evaluations are quantitative. That is, it is concerned with saying something about the order of magnitude of the effects of the project. Of course, qualitative analysis is important, and in most cases is a prelude to quantification — there is little point quantifying if you don’t understand what you are talking about. Quantification, however, provides a discipline and focus for qualitative speculation and provides an important extra dimension when comparing the effects of different projects. When quantifying, there are many sensible approaches that could be adopted. We will focus here on approaches that are broadly consistent with the current approaches to benefit–cost analysis and that could readily be used to augment those approaches. The report begins by reviewing some basic notions of poverty (Chapter 2) and then goes on (Chapter 3) to discuss in principle the ways that agricultural research could influence poverty. Chapter 4 explains, with the use of some examples, a range of analytical approaches that could be taken, and Chapter 5 draws some specific implications for ACIAR.