After summarising the weaknesses of official construction based data, this paper provides forecasts by estimating models of office and retail supply in the UK and some of its regions using quarterly data. It shows the sensitivity of the forecasts to the time period used in the modelling. This is argued to be the result of data problems and the actual history of property supply over the past quarter of a century. Forecasts consequently can vary significantly, depending on the time series used. Suggestions are made as to the best forecasts that can be made using such data.