Survey under-reporting of means-tested programs is well-documented. This paper investigates the degree of under-reporting in the Supplemental Nutrition Assistance Program (SNAP) and Temporary Aid for Needy Families (TANF) programs for the California sample of the American Community Survey (ACS), survey years 2010 through 2012. Previous research has shown that SNAP and TANF have particularly high false negative rates in Census household surveys, but there is much less evidence about the ACS. To assess under-reporting in the ACS, we compare self-reports aggregated to cells defined by geographic location and by demographic characteristics to detailed California administrative counts of participation. Our preliminary estimates yield 2011 ACS under-reporting rates of 51 percent for TANF and 34 percent for SNAP in California, although it varies within demographic groups and across counties. We conduct our main analysis in the absence of one-to-one linkages between our state administrative data source and the ACS. We then provide a correction for under-reporting and assess the extent to which this correction can be applied using solely publicly available data. In this sense, we develop several tools for other researcher to explore when merging surveys with administrative records is infeasible or would result in unworkable delays.