In 2005 India introduced an ambitious national antipoverty program, now called the Mahatma Gandhi National Rural Employment Guarantee Scheme. The program offers up to 100 days of unskilled manual labor per year on public works projects for any rural household member who wants such work at the stipulated minimum wage rate. The aim is to dramatically reduce poverty by providing extra earnings for poor families, as well as empowerment and insurance. If the program worked in practice the way it is designed, then anyone who wanted work on the scheme would get it. However, analysis of data from India’s National Sample Survey for 2009/10 reveals considerable un-met demand for work in all states. The authors confirm expectations that poorer families tend to have more demand for work on the scheme, and that (despite the un-met demand) the self-targeting mechanism allows it to reach relatively poor families and backward castes. The extent of the un-met demand is greater in the poorest states—ironically where the scheme is needed most. Labor-market responses to the scheme are likely to be weak. The scheme is attracting poor women into the workforce, although the local-level rationing processes favor men.