The purpose of this study is to assess empirically the effect of risk on investment decisions for the U.S. manufacturing industries for the period 1962:I-1983:IV. An empirical model describing the interaction between risk and net investment is developed at the outset of this study. This process is broken down into two separate stages. First, an expression for the desired capital stock is derived within the context of the theory of the firm where a representative firm acts to maximize the market value of its equity under uncertainty. Next, a relationship between net investment and changes in the desired capital stock as derived in the first stage is developed. The parameters of the empirical model are then estimated by using quarterly data. Separate net investment functions are fitted for equipment and for structures. Since risk is not directly observable, three proxies are employed. The first proxy is the covariance between output price and the nominal rate of return on the market portfolio. The total nominal rate of return on the New York Stock Exchange (NYSE) is used as the nominal rate of return on the market portfolio. Two other proxies for risk are the yield spread between BAA and AAA corporate bonds and the sample variance of the total nominal rate of return on the NYSE. The impact of risk, regardless of the proxy used, on net investment in equipment and in structures is found to be negative and statistically significant. Net investment models estimated with the first risk proxy are then employed to simulate the response of net investment to the passage of the ERTA of 1981 and the TEFRA of 1982. The impact of risk per se on net investment as well as the extent to which risk affects the effectiveness of above tax policies to either stimulate or deter net investment are also examined. These simulation exercises support the claim that the above tax acts exert significant impact on net investment at the aggregate manufacturing level. However, increased risk is found to mitigate the effectiveness of these tax policies.