Transportation is a key element in the movement of agricultural products to market. The level and nature of demand for different modes has a significant influence on the prices agricultural shippers pay, the routes that are taken, and whether public and/or private investments are warranted. Modal demands, however, are influenced by the geography of the shippers and characteristics of the commodity transported. Shippers located a long distance from the river tend to ship by rail (at least for long-haul movements), while shippers located near the waterway typically ship by barge, as truck-to-barge rates are often considerably lower than rail. The prices paid by shippers also depend on these same characteristics. While truck and barge markets are often thought to be competitive, railroads have considerable latitude in the rates they charge, and the range of observed prices is quite large. In areas where barge is a viable option, railroads may be constrained by barge rates, while in areas where barge is not a feasible option, railroads may choose higher prices (Anderson and Wilson (2008), Burton (1993), MacDonald (1987; 1989)). In between, railroads can choose a set of prices that exclude rail to barge routings in favor of all-rail routings even though the multimodal routing is less costly (Burton and Wilson (2006)). At the root of competition is the presence and availability of substitutes. As noted, in the case of rail and barge transportation, barge is relatively more attractive for locations close to the river. However, as the distance from the waterway increases, barge becomes less attractive and eventually is not used. However, the catchment area for barge—the range at which barge is a feasible mode for shippers—has received relatively little attention, and the relationship between modal demands and catchment areas has typically not been investigated. These relationships are central not only to pricing but also to the feasibility of both public and private infrastructure investments. For example, the U.S. Army Corps of Engineers (USACE) uses barge demand models in their calculation of the benefits of waterway investments. Controversies over their planning models in the late 1990s led to a number of transportation demand studies intending to estimate barge volumes for their calculations. Most of these studies relied on survey data of shippers in a river system (e.g., Columbia-Snake river valleys, Upper Mississippi-Illinois river valleys, and the Ohio). From the surveys, data were collected on shipper choices and options and analyzed using choice methods to estimate demands. Generally, the spatial environment of shippers in these studies was reflected in the choices made. That is, shippers located a long distance from the waterway typically chose rail, while shippers near the waterway typically chose barge. In this sense, the spatial boundaries are inherent in the choices made by the shippers and not used to define spatial boundaries (i.e., the catchment area for barge). The overriding objective of this research is to accurately capture the linkages between the demands for barge and rail freight movement. We develop a methodology—using data for corn, the top agricultural product in terms of tonnage—that considers a wide range of catchment areas using non-survey data. Essentially, we use available data within an area and estimate barge versus rail demand choices. Our approach utilizes the unmasked confidential waybill sample (CWS) of the Surface Transportation Board (STB) and the Waterborne Commerce Statistics (WCS) of the Army Corps of Engineers. The former gives the origin and destination of rail shipments, while the latter gives the origin and destination of barge movements. While this paper focuses on corn, such a methodology could be used to investigate the linkage between modes for any commodity. The main results of this research are: • Most barge shipments terminate in the Central Gulf, while rail shipments are much more diverse, with most terminating in the Illinois, Lower Ohio, Lower Mississippi, or Central Gulf regions. • In the region of study, barge shipments comprise approximately 86 percent of annual tonnage on average. And, the fraction of tonnage shipped by each mode is consistent throughout the sample period (2000 to 2017). • There is uniformity in the results for zones in which both rail and barge are present. That is, there is a preference for barge over rail, holding rates constant, and the rates of barge and rail have an effect that is both economically and statistically important in explaining a shipper’s destination and mode choice. • The preference for barge, however, dissipates, as the distance to the nearest waterway increases. That is, as the distance band (the distance from the river) increases, the coefficient on barge falls. This means that barge is less preferable to rail, given all else is the same, as distance to barge increases. The preference for barge is quite high within 50 miles of the waterway and falls to zero (statistically) for distance bands of about 175 miles. This means that shippers located near the river have a preference for barge, but this preference becomes less important with the distance from the waterway, and by 175 miles or so there is no preference for barge over rail. • As would be expected, the coefficient on rates is negative and statistically significant for all distance bands considered. That is, regardless of the distance from the waterway, we find that a higher rate for a particular mode and route reduces the likelihood a shipper will choose that option. While there are some differences in the coefficient estimates across different distance bands, they are remarkably consistent overall. • Conditional on selecting where to ship, the probability of shipping by barge declines as the barge rate increases but the choice of where to ship does not respond strongly to changes in freight rates, as most of the annual tonnage flows to the Central Gulf. Our results suggest modal substitution for corn is present and persists over a range of different distance bands. The findings provide agricultural stakeholders with information on how the pricing and availability of one mode will impact the other. It also provides an alternative approach to estimating the demands for waterway traffic that both recognizes the effects of competing modes and can be applied to evaluating the effects of waterway proximity on railroad competition and pricing.