This paper examines sub-state spatial and temporal variation in survey misreporting about participation in the Supplemental Nutrition Assistance Program (SNAP). I link several years of the American Community Survey to SNAP administrative records from New York (2008–2010) and Texas (2006–2009) and calculate county false-negative (FN) rates for each year. I find that, within a given state and year, there is substantial heterogeneity in FN rates across counties. In addition, I find evidence that FN rates persist over time within counties. This persistence in FN rates is strongest among more populous counties, suggesting that when noise from sampling variation is not an issue, some counties have consistently high FN rates while others have consistently low FN rates. This finding is important for understanding how misreporting might bias estimates of sub-state SNAP participation rates, changes in those participation rates, and effects of program participation.