We study the effects of increasing the statutory retirement age (SRA) in the Netherlands, using regression discontinuity design and administrative data on the universe of the population. We find clear and large employment effects of the reform. A simple model in which individuals stay longer in their pre-SRA labor market state predicts the treatment effects well. The employment level before the SRA and the retirement hazard at the SRA are the key determinants of the effects of the policy change. Exploring potential explanations for the high hazard rate observed in the Netherlands, our results point to an important role for employers’ effects. (JEL H55, J14, J26, J32)