For children seated next to the struck side, real-world crash outcome was determined for the rear-seat of passenger vehicles over the entire range of side impact crash severities. The method was first to calculate the actual risk for an occupant based on field data. The data sources were non-rollover, tow-away crashes from the 1997 - 2009 National Automotive Sampling System. By limiting the struck passenger vehicle to model year 1985 or newer, field data were identified for a total of 588 children. In all crashes, the child was seated in the rear-seat area on the struck side of the passenger vehicle. A matrix of MADYMO model simulations calculated the response of child dummies over the entire range of the field data. Age-dependent, moderate-to-serious (AIS ≥ 2) injury risk curves were derived and evaluated for children in side impact. Risks to the children were calculated by combining the derived child risk curves with the MADYMO model simulations. The evaluations were conducted from the point of view of an aggregate approximation of AIS ≥ 2 injury rates across the crash severity range for the entire child group of all ages. The results were that the simulations and biomechanical injury risks reasonably matched the actual NASS-CDS-based trauma risks. For all children, the accumulated injury risk approximation was 10.66% and the field-based approximation was 10.73%. This field-based methodology focuses occupant safety design toward a more far-reaching system approach where the entire range of side-impact crash severities and occupant variability are considered.