Research on freight transportation has seen a tremendous increase in the last decades, yet it still lags behind that on passenger travel, particularly at a macro-level suitable for nation-wide policy analysis. A key challenge in freight demand modeling is the availability of data on key drivers of demand - such as cost, time, and trip length - which usually is proprietary and expensive. Moreover freight data available to the public is usually heterogeneous and published by a number of different bodies. In this study we integrate many publicly-available datasets on these attributes into a consistent database describing freight movement at the U.S. level. We then use this dataset to estimate a discrete-choice model of the shares of major single modes - truck, rail, and air, and compare our results with other similar exercises from the transportation economics literature. We thus present an analysis of the effect of generalized transportation costs and infrastructure quality - captured by travel time - on modal split at the national level. We conclude with recommendations regarding freight transportation data that arise from the insights offered by this exercise for policy-makers and practitioners.