This study, which covers New York City in the 1960s and early 1970s, focuses on a previously neglected dimension of welfare — the behavior of cases and groups of cases over time. It counts the cases receiving welfare during different periods and gives the total time and number of times they were on welfare. It measures the effect of case characteristics (payee age and employability, number of children, nonwelfare income) on case behavior and of welfare system, economic, and demographic factors on caseload change. Its investigation of welfare closing, opening, transfer, and turnover rates suggests, among other things, that cases tend to become increasingly dependent on welfare the longer they stay on it. A model that can help explain current caseload change and make conditional predictions of future change is included; its projections are useful in preparing welfare budgets and planning welfare programs.