The World Bank and leading donors have emphasized the importance of evidence-based policy making in addressing entrenched poverty. Adoption of Poverty Reduction Strategy Papers has also required establishing and updating poverty baselines, typically through Living Standards Measurement Surveys. Although the conceptual basis of these surveys has been questioned, little attention has been paid to whether frequent revisions to the longitudinal record have undermined their value. This article argues that changes to the poverty record have been sufficiently frequent and often sufficiently large to compromise the creation of a sound longitudinal poverty record. The lack of transparency also weakens national debate and ownership and undermines the construction of evidence-based policy.