This paper studies the productivity of public investment within the context of a neoclassical growth model. We use recently developed estimates of public infrastructure from Crihfield-Panggabean, Holtz-Eakin, Munnell, and the National Cooperative Highway Research Program (NCHRP) to calculate public and private-sector investment rates. These infrastructure measures are combined with disaggregated data for the states and 282 metropolitan areas in order to estimate the growth model. The estimations show that public infrastructure, however measured, has at most a modest effect on factor markets, and an even smaller impact on growth in per-capita income. Public infrastructure surely plays an important role in metropolitan economies. However, its marginal contribution is no more than, and may be less than, other forms of investment.