This article reviews standard sources of demographic data-censuses, intercensal surveys, registration systems, and specialized surveys and describes their inability to provide accurate data on immigrants, particularly those without documents. We discern a need for data that can identify undocumented migrants and their characteristics, measure trends over time, support longitudinal research, compare the characteristics of migrants before and after they enter, provide sufficient sample sizes for detailed analyses, study transitions between different legal statuses and movements back and forth, and monitor the effects of policy changes on a timely basis. We suggest that the ethnosurvey design satisfies these criteria. We describe the theory, structure, and organization of the ethnosurvey and then describe its application in the Mexican Migration Project. We then highlight its application in other locations around the world and outline an agenda for future comparative research.