Utilizing recently deflated and reconciled U.S. Department of Commerce input-output data from 1947, 1958, 1961, 1963, and 1966, this paper presents an empirical test of the accuracy of input-output forecasts in current dollars and in constant dollars. While the results vary considerably among industries, in general constant-dollar forecasts are more accurate in the long run while in the short run the results are mixed. The implications of these findings for input-output forecasting and research are discussed.