The risk of injury in a front crash depends on both severity and impact configuration, but the accuracy of crush-based estimates of the former depends on the latter. In the current study, output from event data recorders (EDRs) was used to establish estimates of configuration-specific severity and real-world injury risk in crashes similar to consumer test programs. NASS-CDS was queried for all front crashes of vehicles with good IIHS 40% overlap crash test ratings in which the driver was restrained by a belt and deployed front airbag. Vehicles with a complete EDR crash pulse (n=764) or crush-based delta-V calculated from vehicle-specific stiffness values (n=1902) were assigned a crash configuration through photographic review. Cases with delta-Vs from both methods were used to establish configuration-specific EDR-equivalent delta-Vs. The effects of various crash, vehicle, and occupant factors on injury risk were estimated using logistic regression. For large overlaps, moderate overlaps, and center impact crashes, driver age and delta-V had strong effects on the risk of sustaining an MAIS≥3 non-extremity or fatal injury. Vehicles with good ratings in the IIHS moderate overlap test still allow high risk of serious injury in real-world crashes with severities similar to the crash test, especially to older drivers.