The determinants of interstate net migration rates are examined based on the 1992-1993 data for 48 contiguous states and Washington, DC. The weighted least squares (WLS) method assuming dependent variable heteroskedasticity is employed in empirical estimation. A major new finding is that net migration rates and welfare spending per capita exhibit a quadratic or bell-shaped relationship, being positive when welfare spending is relatively low and negative when it is relatively high. The policy implication is that states with relatively high welfare spending such as Washington, DC, New York, New Hampshire, Massachusetts, and others need to re-examine their welfare policy in order not to lose population. Current net migration is also affected by past migration due to information flows and assistance from relatives and friends at the destination place. Other findings indicate that better quality of public education, lower crime, lower tax burden, more job growth, higher wage rates, and better climate are expected to increase net migration.