A number of previous studies have attempted to determine the effect of federal support of research and development (R & D) performed in industry on the rate of private investment in R & D, by estimating regressions of private R & D expenditure on federal industrial R & D expenditure, controlling for total demand (sales) and in some cases other variables. This paper presents arguments and econometric evidence in support of the hypothesis that, due to misspecification of the private R & D equation (i.e., failure to distinguish government sales from other sales), previous estimates of the effect of federal industrial R & D on private R & D funding are seriously upwardly biased.