It is claimed that investment anticipations are very useful in short-term forecasting for predicting actual investment expenditures. This hypothesis is examined and the predictive record of the anticipations for the period 1952 to 1962 is compared to several naive models, a function of endogenous variables only, and a combination of endogenous and anticipatory variables. The authors find that anticipations predict better than naive models in general, but not at turning points. In the manufacturing sector the inclusion of anticipatory variables improves predictions made with a function composed only of endogenous variables, but the inclusion of anticipations does not improve similar predictions in the non-manufacturing sector.