This paper examines the causes of chronic poverty in the remote rural areas (RRAs) of Zimbabwe's semi-arid communal lands. We found that persistent poverty was strongly associated with the structural poverty of Zimbabwe's semi-arid communal areas. Relative urban proximity (non-remoteness) assisted income diversification and improvement in a very poor, socially and politically excluded area. Less excluded but remote areas remained poor but not as poor as those in the excluded area. Livelihoods changed and diversified more in the non-remote area, speeding poverty reduction as measured by an index of perceived change. The paper explores what can be done to improve links between policy-makers and programme designers and poor farm households at risk from drought and examines whether diversification into non-farming occupations offers an exit route from poverty. The paper suggests that Zimbabwe's poor are substantially located in the semi-arid regions. These areas have been neglected by both the state and the private sector. Our findings indicate a massive decline in well-being and consumption between 1993 and 1998 and a failure to rebuild assets after the devastating 1991 drought. We found that semi-arid economies are largely unmonetised as huge swathes of semi-arid Zimbabwe have retreated into subsistence. Markets are not generally sufficiently organised or attractive to engage poor people: barter dominates as a form of exchange and poor households make few cash-based transactions through the market. Key findings are that: The very poorest households have very limited non-farm or off-farm livelihood activities, indicating that for them, improvements in their incomes from agriculture are crucial if their well-being is to improve - either that or new low skill employment opportunities Contrary to expectations for an economy dominated by subsistence production, households with large numbers of people in them as well as those with large numbers of economically active adults are more likely to be poor. Despite attempts to provide a safety net during the 1991-92 drought government grain loans and food-for-work programmes were too limited to be effective. Many poorer households have land lying fallow due to insufficient draught oxen and labour shortages - primarily due to migration and HIV/Aids. The livelihood portfolios which generated the best recovery from drought were waged, nonfarm and mixed farm and non-farm with proximity to urban areas. For most households the value of retained output was more significant than cash income from crop/livestock sales, by factors of between 3 and 10. Lastly we suggest a number of areas for pro-poor policy intervention, including social protection and a focus on improved delivery of social services, (appropriate) agricultural extension and pro-poor financial sector reform.