Since the seminal work of Blanchard and Katz, it has been widely believed that interstate migration causes state-level employment rates in the United States to revert rapidly to normal following a regional employment shock. This paper identifies two sources of bias in conventional estimates of the dynamics of regional labor markets: small sample bias stemming from the use of short time series, and measurement error in survey based series for employment status at the state level. Estimates that use more reliable series and correct for these biases suggest little or no mean reversion in state-level employment rates. Thus the perception that U.S. regional labor markets are highly flexible appears to be incorrect.