The authors analyze the degree of persistence of the unemployment rates of the ten Canadian provinces using quarterly data for the period 1976:1-2005:4. They apply a two-break minimum Lagrange Multiplier (LM) unit root statistic, which, unlike standard unit root statistics (without or with breaks), makes it possible to find the stationarity of the different unemployment rates, giving support to the theory of the natural rate. The authors use the methodology of Bai and Perron (1998, 2003) to estimate a linear model with multiple structural changes to estimate the different degrees of persistence over the different regimes. The results suggest that the degree of persistence decreases when multiple breaks are allowed. Issues regarding the Canadian labor market, the insurance benefits program, interprovincial transfers, and interprovincial mobility are discussed as potential explanations for the results. © The Author(s) 2011.