This paper examines the corporation tax forecasting techniques used by the Institute for Fiscal Studies. For current year forecasts a judgemental forecast is found to have performed better than relying solely on a simple model or information on the receipts available so far in the current financial year. For longer time horizons the judgemental forecast has performed slightly less well than the modelled forecast. While forecasts made later in the financial year have led to more accurate estimates of receipts in the current year no evidence is found that this has improved the accuracy of longer run forecasts. In the short term inaccuracies in the modelling process are found to be more important than errors in forecasting growth in corporate profits. However the latter is still an important component of errors and one that grows substantially in relative importance as the forecast horizon increases.