New data from the Census 2000 estimated that 5.6 million grandparents were living with grandchildren in the United States and a significant percentage of them were responsible for the care of their grandchildren. This paper will compare grandparents response items on Census 2000 and the American Community Survey (ACS) 2000. Data was evaluated to determine the extent to which the ACS was successful in replicating long form Census data for this special population. Three levels of geography were examined: national, state and the 31 ACS counties selected for test sites. The response items are based on questions answered by the population who were age 30 and over in 2000. Respondents were asked if they were living with their own grandchild(ren) under the age of 18. Those who answered Ayes@ were asked if they were responsible for most of the basic needs of their coresident grandchildren and how long they had been responsible for them. Findings suggest that the differences between Census and ACS grandparents’ data seem reasonable and small given sampling error. However, the 90 percent confidence intervals around the ACS estimates are large for smaller areas, making the usefulness of such data be questionable. Several explanations of the differences in estimates based on the two surveys are explored.