According to its advocates, legal titling promises to improve investment (both public and private) and prospects for political order. A persistent puzzle is why its actual impact is often ambiguous or even harmful. Our theory suggests the answer lies in the qualities relating to government: when the state enjoys a monopoly on coercion and administrative capacity, when political decision makers face constraints, and when there are inclusive institutions linking communities to the state, legal titling can improve economic and political well-being. Since these political preconditions for successful legal titling are likely to be absent in conflict-affected states, it is unlikely that legal titling will be effective in such states. The evidence from Afghanistan supports our theory. The legal-titling projects in the country have not worked well, for the reasons just listed. Yet community-based recording of landownership – donor-assisted, community-initiated programs that partner with communities, not the state, to document who owns what land, buildings, and commons – have improved household land-tenure security. These community-based programs illustrate how working with customary governance institutions improves the impact of development assistance.