Previous research has shown that the surveys undercount Medicaid participation and overcount the number of uninsured (Klerman et al. 2009). This study examines the extent of an undercount of Medicaid participants and the subsequent impact on the percent uninsured in the 2006 ACS Content Test. The Medicaid Statistical Information System (MSIS) is an administrative database for the Medicaid program. The American Community Survey tested new questions on health insurance. The MSIS database is linked to the persons in the 2006 ACS Content Test. This research shows that the count of Medicaid participants is underestimated in the 2006 ACS Content Test and the uninsured rate should be lower, assuming the administrative records are without error. The size of the Medicaid undercount for children was larger than expected.