Research on welfare dependency was motivated by the problem of predicting caseloads and analyzing duration and frequency of dependency. This required a model of case flows using aggregate time-series data that incorporated normal movement of cases in and out of dependency, and administrative and economic factors that caused sharp changes in caseload levels. A compilation of case histories from 1967-1972 made it evident that almost two-thirds of monthly welfare support goes to cases continuously on welfare for three or more years. The analysis of half-lives of welfare cases indicate the group receiving ADC for two and more children had a half-life no longer than 2.5 years, and other groups have shorter half-lives. Thus the idea of a permanent welfare population is not accurate. Based on the data available, it seems that welfare cases become increasingly dependent on welfare the longer they stay on welfare. 32 pp. Ref.