High‐rate vertical acceleration and resulting force measurements from underbody blast events are routinely being replicated in laboratory environments. A knowledge of the input force time history requires acceleration time history data of the rig structure designed to measure force for mass compensation computations. These mass compensation computations are difficult to perform, however, as a function of accelerometer resonance and the accelerometer's propensity to express modes in the load‐measuring structure. One needs to apply a filter to such accelerometer data, but at this time there is no standard methodology for deriving an appropriate filter. Current researchers trying to measure input forces either do not report the filter frequencies used in mass compensation computations or use current standards derived for different applications than what the experiment may entail. The hypothesis of this study is that, with load cells of appropriate response characteristics, one can infer an optimized filter for acceleration time histories used in mass compensation computations. The proposed methodology applies a number of filters to acceleration data, then computes the standard deviation of mass cancelled data with no specimen present, and defines the best fit acceleration data as those with the minimum standard deviation after mass compensation has been performed.