Forecasters are often critised for failing to predict outcomes, and this criticism can be justified if they rely on point estimates in their forecasting rather than providing a measure of the uncertainty around their estimates. There are a number of ways to provide measures of uncertainty, but perhaps the most widely used involves the calculation of past forecast errors and hence the use of an outcome distribution based on those errors. Evaluating forecast uncertainty is particularly important at present as events in US and European financial markets have meant that a significant degree of judgement has to be utilised in order to produce a projection for the next two years. The scale of loses in the US mortgage and related markets are unknown, and the impacts of even an agreed set of losses are uncertain.