I have been part of the Quantitative Research Unit at the Danish national pension fund ATP since 2002. We are a powerhouse business development unit which uses computer science, mathematical modelling and statistical concepts as our main tools. Our work is inter-disciplinary and we have developed, e.g., hedging and investment strategies, risk management frameworks, longevity models and the current pension product. In the talk, I will describe how a pension fund operates and some of the challenges faced due to the economic landscape and regulatory demands. I will give examples of some of the many problems where statistical thinking is useful (and much needed), and I will describe in some detail two of our latest projects.