Expectations about length life and when to retirement are crucial for retirement decisions and consumption and savings decisions prior and post actual retirement. It is fair to say that even decisions regarding health care consumption and health behaviors will be influenced by these expectations. Understanding how expectations evolve when people approach retirement and how unanticipated changes in the institutional environment influence retirement planning and consumption and savings decisions is important for both scientists, pension funds and policy makers. In this presentation we look at the consequences of an unanticipated pension reform for retirement expectations, savings behavior and actual retirement decisions. In the summer of 2005 it was announced that as of 2006 the pension rights of those born in 1950 or later were substantially reduced, while the pension rights of those born in 1949 or earlier were unaffected. More specifically, the pension reform implied for the 1949 cohort (aged 57 in 2006) that they could retire at age 61 at 70% of the last earned wages. For the 1950 cohort the replacement rate was 64% at age 61 and they had to work an additional 13 months to obtain the 70% of their slightly older counterparts. In earlier work (De Grip, Lindeboom & Montizaan, 2012) we compared the mental health status, two years after the reform, of workers born in 1949 and 1950. It was found that this unanticipated exogenous change in the retirement system strongly affected the mental health status of those affected by the reform: in 2008, two years after the reform, depression rates of the 1950 cohort were about 40% higher than among the 1949 cohort. In this presentation we proceed one step further and look at the evolution of retirement expectations after the unanticipated reform and the behavioral responses, i.e. savings behavior and actual retirement behavior. The relatively abrupt change in the environment and the fact that it was announced late in the game makes it difficult for a substantial share of the workers to fully guard themselves against retirement income losses while keeping leisure years (retirement date) unchanged. We link administrative data and survey data to obtain information about retirement expectations and realizations and savings decisions. This allows us to measure income – leisure trade-offs in a situation where workers have to revise their pension planning, induced by policy changes or other aggregate shocks.