The risk of infrastructure investments is driven by unique factors that cannot be well described by standard asset class factor models. We thus create a nine-factor model based on infrastructure-specific risk exposure, i.e., market risk, size, value, momentum, cashflow volatility, leverage, investment growth, term risk, and default risk. We empirically test our model on a large dataset of U.S. infrastructure stocks in different subsectors (utility, telecommunication, and transportation) and over a long period of time (1983 to 2011). The new factor model is able to capture the variation of infrastructure returns better than the Fama/French three-factor, the Carhart four-factor or the extended Fung/Hsieh eight-factor models. Thus, our model helps to improve the evaluation of infrastructure funds and to better determine the cost of capital of infrastructure firms, something that is increasingly relevant in light of the growing need for privately financed infrastructure projects.