One of the greatest challenges faced in the design of realistic occupant protection systems is an accurate statistical model of what is really needed. The paucity of data is this realm hinders designers of standards alike.Ideally, a model of crash statistics would correlate, for significant accident modes, injury level (as measured by AMA Abreviated Injury Scale “AIS”) with some adequate measure of crash intensity. Having this information, not only could the required level of safety design be ascertained, but also the justifiable economic expenditure could be estimated.This paper treats the statistical basis for deployment of a data retrival system. It provides a basis for estimates of the amount of data required, the number of vehicles to be instrumented, the crash severity trigger levels, and the economics of recorder installation, for various levels of injury and fatality. Results are presented in graph of time required to achieve N data points vs. time value of expected data, as applied to current U. S. statistical information.The paper concludes with a brief outline of one feasible system for crash data recording and retrieval.