This paper analyses the determinants of structural unemployment rates in a two-stage approach. First, time-varying NAIRUs are estimated for a panel of OECD economies on the basis of Phillips curve equations using Kalman filter techniques. In a second stage, the estimated NAIRUs are regressed on selected policy and institutional variables. As predicted by theoretical wage-setting/price-setting models, the level of the tax wedge and the user cost of capital are found to be important drivers of structural unemployment. Consistent with earlier studies, the level of product market regulation, union density and the unemployment benefit replacement rate also play an important role in explaining changes in the NAIRU although there is considerable variation in estimates across countries. Nonetheless, the set of structural variables provides a reasonable explanation of NAIRU dynamics over the period 1978-2003, even though recent decreases are better explained than the earlier surge.