The accuracy of velocity determination from common mid-point (CMP) ground penetrating radar surveys has been assessed in the past and found to be sufficient for migration and depth determination. Increasingly, these velocities are also being used to quantify subsurface physical properties such as water content. This paper demonstrates that small errors in measured velocity can result in large errors in these derived properties. We have evaluated the size of some error sources with specific reference to a given glaciological model and common glaciological survey conditions. At worst, large static errors and 3% errors in measured offset result in interval velocity errors of ∼8.6%. This error is large enough that derived water content has an error greater than 80% (e.g., 2.8 ± 2.4 volumetric % water). Common acquisition and processing practices result in ∼4.9% interval velocity errors (corresponding to ∼50% error in water content, e.g., 2.8 ± 1.4 volumetric %); best practices could result in errors as low as ∼0.6% in interval velocity (corresponding to <10% error in water content, e.g., 2.8 ± 0.2 volumetric %), but to achieve this level of precision requires a revision of common practice for CMP acquisition and processing. This revision would need to result in improved assignment of time zero and a consistent definition of reflection event arrival time (to provide travel time errors smaller than 0.5%), and accurate measurement of CMP geometry.