This paper draws together findings from different elements of a research project examining critical components of pro-poor agricultural growth and of policies that can promote such growth in poor rural economies in South Asia and Sub-Saharan Africa. Agricultural growth, a critical driver in poverty reducing growth in many poor agrarian economies in the past, faces many difficulties in today’s poor rural areas in South Asia and Sub-Saharan Africa. Some of these difficulties are endogenous to these areas while others result from broader processes of global change. Active state interventions in ‘kick starting’ markets in 20th century green revolutions suggest that another major difficulty may be current policies which emphasize the benefits of liberalization and state withdrawal but fail to address critical institutional constraints to market and economic development in poor rural areas. This broad hypothesis was tested in an analysis of the returns (in agricultural growth and poverty reduction) to different government spending in India over the last forty years. The results reject the alternate hypothesis underlying much current policy, that fertilizer and credit subsidies, for example, depressed agricultural growth and poverty reduction in the early stages of agricultural transformation. The results show initially high but then declining impacts from fertilizer subsidies; high benefits from investment in roads, education and agricultural R&D during all periods and varying benefits from credit subsidies over four decades; low impacts from power subsidies; and intermediate impacts from irrigation investments. These findings demand a fundamental reassessment of policies espousing state withdrawal from markets in poor agrarian economies. Given widespread state failure in many poor agrarian economies today, particularly in Africa, new thinking is urgently needed to find alternative ways of ‘kick starting’ markets – ways which reduce rent seeking opportunities, promote rather than crowd-out private sector investment, and allow the state to withdraw as economic growth proceeds. To investigate some of the potential opportunities and difficulties in achieving pro-poor agricultural growth in poor rural economies today, empirical work on Malawi and Zimbabwe used farm-household, rural economy and CGE models to analyze the structure of different rural livelihoods and to simulate policy impacts on livelihoods, rural growth and poverty. This work, together with findings from wider reviews and the Indian econometric work, highlighted very diverse constraints, opportunities and behavior among different household types and confirmed the importance of smallholder agricultural growth for poverty reduction through its impacts on labour and grain markets (even where it accounts for less than 50% of rural incomes). However, large productivity increases are needed from labour saving technical change if smallholder agriculture is to drive pro-poor growth, backed up by growth in the rural non-farm economy and longer-term tradable non-agricultural growth drivers for sustained poverty reduction. Where productive labour demanding technologies exist, there are large potential pro-poor growth benefits from reducing transaction costs to improve access to agricultural markets and from increased smallholder household liquidity. Liquidity constraints lead to important synergies between some forms of welfare support and agricultural productivity and growth, while institutional development is needed to improve access to input, output and financial markets. Market intervention policies that stimulate the development of otherwise thin food grain and input markets can stimulate pro-poor growth if the major practical problems in the implementation of these policies can be addressed, and if these policies are backed up by significant and long term, but flexible and targeted investments. Where widespread and large-scale increases in productivity cannot be achieved, agriculture will not be able to drive the overall rural growth and structural change needed to provide the base for significant poverty reduction. It still, however, has a vital role to play in ‘livelihood protection and enhancement’: supporting people’s existing livelihoods and maintaining the natural resource base. Policies and investment in this ‘livelihood protection and enhancement’ role for agriculture should not be so demanding as those required for pro-poor agricultural growth, in terms of minimum scale, impact, coordination or institutional capacity needed for success. Nevertheless, significant investments will still be needed to develop appropriate technologies, to facilitate coordinated service provision for smallholder farmers, and to encourage a more secure and favorable political and economic environment conducive to agricultural and rural investment. The returns to these investments need to be analyzed against the costs of livelihood and natural resource failure in these poor rural areas, and against the human and fiscal costs of these rural communities’ becoming increasingly dependent on long term welfare support and emergency relief.