This paper estimates the aggregate productivity effects of Marshallian externalities generated by foreign direct investments (FDI) in the US. In contrast to earlier work, this paper puts special emphasis on controlling for Marshallian externalities and other intra- and inter-regional spillovers generated by domestic firms. The productivity effects of these externalities may, if not accounted for appropriately, be falsely attributed to FDI. This paper also deals with the potential endogeneity of FDI and the presence of spatial lags by employing a system generalized method of moments (GMM) estimator. We use a regional production function framework that models Marshallian externalities and other intra- and inter-regional spillovers explicitly as determinants of total factor productivity, and tests several empirical specifications of this model, using data for US states from 1977—2003. The results indicate that FDI does, in fact, generate positive externalities, while those from domestic firms are negative.