THE ability of firms to predict the movement of their future sales and other short-term variables is of interest to economists both for macroand micro-economic planning reasons. If firms, in the aggregate, fail to provide fairly accurate predictions of the level, or at least the direction, of change, in these variables, then anticipations surveys, providing series for predicting the behaviour of the economy either directly or by suggesting causal relationships,1 are likely to be misleading. At the micro level, if individual firms are unable to predict satisfactorily their own sales, and other variables, then doubt is cast on theoretical formulations of firm and market behaviour which tend to ignore, or play down, uncertainty. A number of sales and employment surveys have been completed in the United States2 and elsewhere3 so that a body of evidence has accumulated on the short-term predictive accuracy of such aggregated data and some tentative analyses have been made. However, little attention has been devoted to measuring the accuracy of individual company forecasts. Two questions are important: how accurately do firms anticipate