In this paper we study the regional evolution of unemployment in Argentina for the period 1980-1997. We show that its regional unemployment structure is not very persistent. We then apply panel data unit root tests to show that unemployment itself is not highly persistent. Indeed, we reject the null of a unit root in the unemployment series. Finally, we model the conditional means of regional unemployment. We adopt a dynamic two way fixed effects error component model specification. We measure the persistence of unemployment to shocks based on our conditional model. We believe this provides us with a better measure of persistence than the commonly used in the literature. We find a low degree of unemployment persistence to shocks. Finally, we also find regional factors that explain regional unemployment differences and whose changes may account for the low persistence of the regional unemployment structure.