Abstract: Critical Illness (CI) insurance is a form of long term insurance which pays a sum insured on the diagnosis, within the terms of the policy, of one of a specified list of illnesses, for example cancer or a heart attack. It is a very popular type of insurance in the UK. At its peak, in 2002, over one million new policies were sold. In this talk I will give an overview of a recent project which used data from 1999 - 2005 covering nearly one half of the UK market, the objective being to develop and parameterise a probabilistic model which can be used to price and value CI policies. The talk will cover technical aspects - Markov models, parameter estimation, variable selection - and the practical difficulties which inevitably arise with any project involving a large data set. Results will be presented.