In an influential study, Nickell et al. (2005) find that institutions explain most of the variation in OECD unemployment, using panel data for 20 countries from 1960 to 1995. However, the importance of Nickell et al. (2005)’s conclusions has spurred a lively debate, and several authors have criticized their findings. This paper re-assesses the main findings in Nickell et al. (2005), benefitting from the inclusion of twelve additional years of data. A dynamic simulation of their main unemployment equation shows that unemployment is severely underpredicted in the post sample period for 17 of the 20 countries, while it is only overpredicted for one country. The analysis shows that the underprediction is largely driven by the model dynamics and that the accounting exercise in Nickell et al. (2005) is not suited to analyze how much of the variation in unemployment that can be attributed to changes in institutions in the post-sample period. However, there is a clear tendency that countries which have changed their institutions in an ”employment-friendly” way, like Denmark and Finland have experienced a larger reduction in unemployment than the countries that have changed their institutions in the opposite direction like Germany and Portugal. This indicates that labour market institutions affect unemployment in the direction found by Nickell et al. (2005). ∗I would like to thank Erik Biørn, Steinar Holden and Ragnar Nymoen for comments and discussion. The numerical results in this paper were obtained by use of Stata 9. This paper is part of the project Demand, unemployment and inflation financed by the The Research Council of Norway.