Planning and development for Census 2000 is well underway. A fundamental strategy to achieve the goals of increased accuracy at reduced cost is the use of innovative statistical sampling methods as an integral part of census data collection. Census 2000 will be the first decennial census that will use sampling to introduce efficiencies into the follow-up of households that do not mail back a questionnaire. Census 2000 will also include, for the first time, an integrated coverage measurement survey to ensure a complete enumeration by correcting the initial census results for persons who would otherwise be missed. The Census Bureau has been conducting research throughout the decade to develop these statistical methodologies, leading to a complete plan for the implementation of these methodologies. The research has been developed and discussed with the a number of advisory and stakeholder groups including the National Academy of Sciences, the American Statistical Society, and the U.S. Congress.

As was the case for previous censuses, the incorporation of sampling as an integral part of census data collection is an extremely controversial topic. A number of issues, both technical and political, have been raised and extensively debated. This paper will discuss the results of the research that has been conducted this decade and will present the final plan for implementing statistical sampling in Census 2000. The paper will also address the key methodological decisions that have been made in developing the methodologies, and will present some preliminary findings from the 1998 dress rehearsal. Finally, the issues that have been raised will be discussed.