The vast majority of Individual Retirement Account contributions represent net new saving, based on evidence from the quarterly Consumer Expenditure Surveys (CES). The results are based on analysis of the relationship between IRA contributions and other financial asset saving. The data show almost no substitution of IRAs for other saving. While the core of the paper is based on cross-section analysis, important use is made of the CES panel of independent cross-sections that span the period during which IRAs were introduced. Estimates for the post 1982 period, when IRAs were available to all employees, are based on a flexible constrained optimization model, with the IRA limit the principle constraint. The implications of this model for saving in the absence of the IRA option match very closely the actual non-IRA financial asset saving behavior prior to 1982. IRA saving does not show up as other financial asset saving in the pre-IRA period.