This research compares the likelihood of exiting TANF with and without employment and the effects of important state TANF rules on welfare exits in more disadvantaged (large Rustbelt cities and poor southern nonmetro) and less disadvantaged (other metro and other nonmetro) areas during the 1996-2003 post-welfare reform period. Hierarchical competing risk models using individual-level data from the 1996-99 and 2001-03 Panels of the Survey of Income and Program Participation merged with state-level data from various sources show that female TANF participants in poor southern nonmetro areas are the least likely to exit TANF with work, and participants in large Rustbelt cities are less likely to exit TANF with work than those in other metro areas. Non-work TANF exits are more likely to occur in other nonmetro areas than in other metro areas. Importantly, the effects of state welfare rules on TANF exits differ across places of residence. For example, stringent time limit policies promote work exits in large Rustbelt cities but promote non-work exits in poor southern nonmetro areas. More lenient earned income disregards are significantly related to remaining on TANF in poor southern nonmetro areas but promote work exits in all other places. Findings from this paper imply that states should not take a "one-size-fits-all approach" to reduce welfare caseloads. (Contains 3 tables and 3 footnotes.)