Unemployment compensation (UC), a key part of the social safety net, is not individually reported in the American Community Survey (ACS), an issue for data users wishing to estimate the impact of UC on those in need. Using regression techniques, we predict the share of reported "other" income attributable to UC in the 2008 ACS data for Georgia, Illinois and Massachusetts. We review UC underreporting in the ACS data and discuss one possible method of correction using the Transfer Income Model, Version 3 (TRIM3). We demonstrate the effect of the UC underreporting correction on the poverty rate and poverty gap.