I document a new empirical pattern of internal mobility in the United States. Namely, county-to-county migration and commuting drop off discretely at state borders. People are three times as likely to move to a county 15 miles away, but in the same state, than to move to an equally distant county in a different state. These gaps remain even among neighboring counties or counties in the same commuting zone. Standard economic explanations, which emphasize differences in utility or moving costs, have little explanatory power. Cross-border differences in observables, amenities, state occupational licensing, taxes, or transfer program generosity do not explain this border effect. However, county-to-county social connectedness (as measured by the number of Facebook linkages) follows a similar pattern, and there is suggestive evidence that this is driven by a so-called "home state bias," rather than alternative explanations such as information frictions or network ties. I show that this reluctance to cross state lines has real economic costs, resulting in local labor markets that are less dynamic after negative economic shocks.