The drug felony lifetime ban on Temporary Assistance for Needy Families (TANF) affects thousands of individuals with felony drug convictions in the United States. Federal law allows states to choose to opt out or modify the full ban. Prior research has treated the ban as a binary outcome, characterizing anything but a full ban as a sign of state reform of this harsh collateral consequence. We argue that modified versions of the ban, which simultaneously allow greater access to public aid while also monitoring and sanctioning recipient behavior, have been overlooked but pose important theoretical and empirical challenges to this narrative. To address this gap, we analyze state discretion in the implementation of the drug felony lifetime ban on TANF receipt between 1997 and 2010 utilizing a multilevel multinomial modeling strategy. Results reveal that distinct patterns of state-level factors are associated with each form of the ban, highlighting the need to treat modified bans as unique policy choices in their own right. Our study informs the understanding of state implementation of collateral consequences that straddle both the penal and welfare systems in the United States.