Landholding class has been widely advocated as a targeting indicator for fighting rural poverty in South Asia and elsewhere. Supporters have pointed to the correlation between poverty and lack of land, and have also claimed that land-based targeting raises output. The paper tries to assess the impact on rural poverty in Bangladesh of a benchmark policy under seemingly ideal conditions, generalizing earlier results of Ravallion (1989) on improved data. Poverty is reduced, though the maximum gains turn out to be small, even allowing for pro-poor productivity effects consistent with the farm size-productivity relationship in Bangladesh. The gains are likely to be even smaller under real-world incentive and political economy constraints.