The Marcellus Formation, a Devonian gas shale in the Appalachian Basin, is a heterogeneous rock as the result of a complex depositional, diagenetic, and deformational history. Although it is overpressured over a large portion of its economic area, the origin and distribution of pore pressure within the gas shale are not well-understood. We have used the sonic properties of the Marcellus and statistical analyses to tackle this problem. The sonic data come from a suite of 53 wells including a calibration well in the Appalachian Basin. We first analyze the influence of various extrinsic and intrinsic parameters on sonic velocities with univariate regression analyses. The sonic velocities of the Marcellus in the calibration well generally decrease with an increase in gamma-ray american petroleum institute (API) and increase with density and effective stress. Basin-wide median sonic velocities generally decrease with an increase in median gamma-ray API and pore pressure and increase with burial depth (equivalent confining stress), effective stress, and median density. Abnormal pore pressure is verified by a stronger correlation between the median sonic properties and effective stress using an effective stress coefficient of approximately 0.7 relative to the correlation between the median sonic properties and depth. The relatively small effective stress coefficient may be related to the fact that natural gas, a “soft” fluid, is responsible for a basin-wide overpressure of the Marcellus. Following the univariate regression analyses, we adopt a multiple linear regression model to predict the median sonic velocities in the Marcellus based on median gamma-ray intensity, median density, thickness of the Marcellus, confining pressure, and an inferred pore pressure. Finally, we predict the pore pressure in the Marcellus based on median sonic velocities, median gamma-ray intensity, median density, thickness of the Marcellus, and confining pressure.