The present study had three objectives: (1) define a reasonable number of categories to bin head injuries, (2) develop an overarching risk function to estimate head-injury probability based on injury probabilities pertaining to those subordinate categories, and (3) assess the fidelity of both the overarching function and approximations to it. To achieve these objectives, we used real-world data from the National Automotive Sampling System (NASS), pertaining to adult drivers in full-engagement frontal crashes. To provide practical value, we factored the proposed US New Car Assessment Program (US NCAP) and the corresponding Request for Comments from the government. Finally, the NASS data stratifications included three levels of injury (AIS1+, AIS2+, AIS3+), two levels of restraint (properly-belted, unbelted), and two eras based on driver-airbag fitment (Older Vehicles, Newer Vehicles).For Objective 1, three categories were studied (“brain-related,” “bone-related,” and/or “otherwise-related”), as well as the overarching category (“head”). The following trends for category-based injury rates (IR) were noted: (a) more than 84% of head-injury cases involved only one of the injury categories, (b) IRAIS3+ < IRAIS2+ < IRAIS1+, (c) IRbelted < IRunbelted, and (d) IRNewerVehicles < IROlderVehicles. We noted otherwise-related injury was primary to AIS1+ studies, but brain-related injury was primary to AIS2+ and AIS3+ studies. Therefore, the AIS3+ head-risk assessment in the proposed US NCAP will likely improve most by improving the estimation of brain-related injury.For Objective 2, a three-category overarching risk function was assumed, based on subordinate injury probabilities: max(Pbrain, Pbone, Pother). For each level of injury (AIS1+, AIS2+, and AIS3+), this function was evaluated for 88 possible bins, and cross-correlated with the direct-estimate Phead. Each resulting regression-line slope nearly equaled 1.0, with R2=0.96 for AIS1+ injury, R2=0.98 for AIS2+ injury, and R2=0.99 for AIS3+ injury.For Objective 3, the methodology used of the second objective was repeated for one- and two-category functions. We noted that, as category count decreased, fidelity degraded. However, the fidelity of the two-category function yielded nearly identical fidelity as the three-category function. Pertinent to the proposed US NCAP, the best-available two-category function for AIS3+ assessment would involve brain- and bone-related assessment.