The passage, in 1996, of the Temporary Assistance for Needy Families (TANF) program has enjoyed extraordinary political support, largely as a result of the research study by the Council of Economic Advisors, which asserted that TANF actually‘works’. However， significant debate has surfaced within academic circles regarding both the theoretical and empirical foundations for such findings. This paper sets out to shed some light on the precipitous decline in welfare caseloads during the period 1996-1999, undertaking comparisons with regard to the variations in caseload reduction and poverty rates by taking into consideration both trend and maturity factors. The study finds that discretion at state level, in respect of both program content and enforcement, has had a significant impact on caseload reductions whilst having little to do with the actual alleviation of poverty. The conclusions of this paper suggest that future longitudinal studies should take into account distributional issues, such as the consequences of policies for different groups, regions or time periods, in order to effectively evaluate the overall impact of the TANF program on mothers and their children.