This thesis aims at modeling the mortality experienced in the portfolio of a pension insurer. Not only is past mortality modeled, also projections about future mortality are made. The globalmodel is split in two parts: a stochastic model for population mortality which allows for forecasts, combined with a point estimate from a portfolio mortality model. Future insurer-specificlife tables are obtained for the period 2010-2060, along with 95% confidence intervals of the death probabilities. Results are compared to other projections used in Dutch (insurance) practice. Theobtained, insurer specific, mortality model is then used to develop an internal model for assessing the longevity risk of the insurer in the Solvency II framework. The results for this model are compared to the results for the standard approach.