Most economists conclude that the U.S. regions have converged in per capita earnings during a majority of the 20th century, though controversy abounds over the methods employed to test for such convergence. Using time-series techniques, this paper finds evi - dence that the U.S. regions have conditionally converged in per capita earnings. The findings in this paper differ from cross-sectional studies, which implicitly assume that all regions converge toward the same steady-state and at the same rate. The findings in this paper differ from other time-series studies with its use of recursive parameter estimates.1