Whilst the preliminary nature of and subsequent revisions to estimates of real output provide a source of data uncertainty to macro-economic forecasters, many macro-economic data (e.g. exchange rates, interest rates and commodity prices) are easily available and not subject to measurement error. In contrast, real estate forecasters are faced with substantial hindsight uncertainty. Hendry and Clements (2003, 303) identify two fundamental sources of forecast uncertainty stating that ìall econometric models are mis-specified, and all economies have been subject to unanticipated shiftsî. This paper focuses on the contribution of data uncertainty to model mis-specification and consequent forecast uncertainty. Using data provided by three major real estate advisory firms, we investigate the level and pattern of variation in the measurement of historic real estate values and market indicators for the main European office centres. We investigate the ways in which the choice of historic time series can influence: i. model specification, ii. estimation of the coefficients of the explanatory terms and iii. the level of forecast disagreement.