The airbag can fire at any given time during a crash. To qualitatively and quantitatively address the probabilistic nature of the airbag fire-time as a result of a crash, a statistical data based model was used. Two models were constructed:​ one was based on rigid barrier impacts and the other used offset deformable barrier impacts. These models were developed previously and some of the preliminary results were presented at the 1998 SAE conference in Detroit. Since that time the models have been refined by inclusion of additional data, and the results of the refined models were compared to the results from the previously constructed models. The models were used to address the effects of raising the “threshold velocity” and the risks of an occupant contacting the airbag module as a result of late firing. Although the individual numerical values have changed, the indicated general trends remain the same. Raising the “threshold velocity” may or may not decrease the number of occupants on the module, depending on the accuracy of NASSJCDS and fire/no-fire as a function of velocity (c-censor*), but it appears to reduce the effectiveness of the airbag when it does fire. The model indicates that decreasing the width of the c-censor, i.e., delta V from no fire to all fire, may be useful in reducing the number of occupants on the module without decreasing the effectiveness of the airbags.