HDR has conducted a feasibility study for potential passenger rail service along the Central Corridor Line, a 100-mile rail corridor that connects Brattleboro, Vermont and New London, Connecticut. Traditionally, four-step regional travel demand models have been utilized for forecasting transit ridership. However, no readily available travel demand models existed for this multi-state corridor study since its study area is not fully covered by any regional model. In addition, limited socio-economic data was available related to employment, which is a critical input to the four step modeling process. As a solution, a direct demand model (DDM) was developed by creatively consuming cellular phone sighting data. The AirSage WiSE (Wireless Signal Extraction) platform was used to collect and analyze wireless network data to determine the location and movement of cell phones, which were subsequently converted to person trip origin-destination matrices. The DDM was used to establish a relationship between existing rail ridership and trip information derived from cellular data in one analogous passenger rail corridor, which was successively applied to the proposed corridor for forecasting future ridership. The underlying assumption in the direct demand modeling approach was that the observed railroad service usage in the existing comparable corridor was an indicator of proposed and future railroad service usages. A multivariate regression analysis was conducted to formulate and calibrate the DDM, in which rail ridership at the station level (the dependent variable) was related to key independent variables such as number of person trips, number of trains operating daily, and other important level-of-service variables that were considered to influence ridership. The estimated regression equation demonstrated an acceptable goodness of fitness and produced reasonable forecasts. Unlike the traditional four-step model which requires significant data and rigorous calibration, the DDM was a more practical alternative, especially in this case where a regional travel demand model was not readily available. Further, use of cellular phone data can be an effective surrogate for population and employment data in corridors where socio-economic datasets are not easy to obtain. The paper discusses the details of model development, results, and aspects of model transferability over geography and time.