In this paper, we have recognized that the relatively constant inequality in the distribution of U.S. family income masks dramatic changes in the structure of the income distribution and the composition of personal incomes. Using relatively new procedures for decomposing the Gini measure of income inequality, we have gained a better understanding of the relationships among changes in the sources of income and the income distribution. This is facilitated through the use of data collected by the Census Bureau in the Survey of Income and Program Participation, which is the only set of data currently available that contains exhaustive lists of income and asset information at the household level. The empirical results are used in conjunction with data on transfer program expenditures to gain some perspective on the relative costs of reducing income inequality by increasing program benefits.