This paper proposes an empirical method for estimating a long-run trend for the unemployment rate that is grounded in the modern theory of unemployment. I write down an unobserved-components model and identify the cyclical and trend components of the underlying unemployment flows, which in turn imply a timevarying estimate of the unemployment trend, the natural rate. I identify a sharp decline in the outflow rate—the job finding rate—since 2000, which was partly offset by the secular decline in the inflow rate—the separation rate—since the 1980s, implying a relatively stable natural rate, currently at 6 percent. Numerical examples show that slower labor reallocation, along with the weak output growth, explains most of the persistence in unemployment since the Great Recession. ; Contrary to the business-cycle movements of the unemployment rate, a significant fraction of the low-frequency variation can be accounted for by changes in the trend of the inflows, especially prior to 1985. Finally, I highlight several desirable features of this natural rate concept that makes it a better measure than traditional counterparts. These include statistical precision, the significance of required revisions to past estimates with subsequent data additions, policy relevance and its tight link with the theory.