Previous studies have suggested injury assessment reference values (IARVs) for lower neck injury based on scaled upper neck values. This study developed independent flexion and extension IARVs for the lower neck by matching Anthropomorphic Test Device (ATD) data to impact-tested post-mortem human subjects (PMHSs) with mid- to low-cervical spine injuries. Pendulum and sled tests with Hybrid III midsize male and small female ATDs were run under conditions mimicking those of published PMHS torso drop-sled tests and other PMHS studies. Measurements included upper and lower neck forces and moments, head acceleration, head rotation rate, and head/neck angles for the pendulum tests. Rear impacts corresponding to rigid seatback tests without a head restraint produced lower neck extension moments that increased dramatically with test severity, as measured by increasing delta-V and/or decreasing pulse duration. In contrast, upper neck extension moments increased only modestly with test severity, and all recorded peak values of upper neck force and moment, as well as calculated Nij, were below currently accepted IARVs. Our proposed lower neck IARVs are well below those previously published, which are based on simple scaling of upper neck IARVs. Frontal impact tests provided an indication that injuries may occur at lower neck flexion moment values that are below suggested tolerances; however, since the available PMHS dataset only included injured subjects for which the observations were left censored, no IARV could be determined. Relationships between peak moments at the upper and lower neck, as well as between male and female ATDs in tests of matching severity, were examined and compared to standard scaling used for previously proposed IARVs. Peak moment versus tension and compression forces were also evaluated, yielding additional data points for estimating critical intercepts for lower neck injury criteria (Nij).