According to data from the 2000 Decennial Census and the 2005-07 American Community Survey, household overcrowding has decreased drastically between those two periods, declining by over half in some cases. This change would indicate either an equally dramatic improvement in housing conditions or a serious incompatibility in data from the two sources. This technical brief examines several issues by reviewing changes in the two factors that are used to calculate persons per room, and analyzes their relative contribution to the apparent decline in overcrowding. The decrease can be attributed to observed changes in the distributions of housing units and households by size, with the former playing a much larger role. The brief concludes with speculations about possible data problems due to changes in the survey methods used in 2000 and in 2005-2007.