As is well known, the U. S. national accounts contain a “Statistical Discrepancy” which serves to reconcile estimates ofGNPfrom the product and from the income side. While the Discrepancy is a measure only of overall consistency rather than of error, its magnitude and pattern over time do give some indication of the quality of the national accounts data. In the past, the Statistical Discrepancy has not always behaved as a quantitatively unimportant random error term—a disturbing sign with regard to theGNPstatistics. The presence of a systematic Discrepancy also causes difficulties in some specific problems. For example, it is necessary to assume a value for the Discrepancy in the identities of an econometric model when forecasts are made. The Discrepancy can also seriously affect estimates involving thoseGNPcategories, such as savings, which are customarily calculated as residuals with the consequence that the Discrepancy is allocated to them.