Economic growth has been low and the incidence and numbers of poor people remain very high in some parts of the world, notably in sub Saharan Africa and some parts of South Asia. Projections for poverty reduction suggest that these regions are likely to continue to hold very large numbers of very poor rural people in the foreseeable future. Theoretical arguments and empirical evidence suggest that in poor agrarian economies both the processes of structural change within national economies and micro-economic relations within rural economies give agriculture (and particularly intensive cereal based growth) a pre-eminent and unique role in economic development and in poverty reduction. However, reliance on pro-poor agricultural growth as the main weapon against rural poverty today faces more difficult challenges than those faced in the green revolution areas in the latter part of the 20th century, due to a number of features that together increase risk and uncertainty and raise costs and/or lower returns to agricultural investment. Many of these difficulties are endogenous to today's poor rural areas, others result from broader processes of global change, but it is argued that some are the direct result of policies supporting liberalisation and withdrawal of the state. A review of the green revolutions of the 20th century suggests that state interventions in agricultural markets were widely used and important in supporting sometimes short periods of critical market and technological development in the process of rural growth. Unfortunately the benefits of such interventions have been overlooked as a result of their very evident inefficiency and high costs, without a clear understanding of their institutional benefits. Policy and research implications of this analysis are discussed.