The upper and lower neck loads in the restrained Hybrid III dummy and Test Device for Human Occupant Restraint (THOR) were computed in simulated frontal impact sled tests at low, medium, and high velocities;​ repeatability performance of the two dummies were evaluated at all energy inputs;​ peak forces and moments were compared with computed loads at the occipital condyles and cervical-thoracic junctions from tests using post mortem human surrogates (PMHS). A custom sled buck was used to position the surrogates. Repeated tests were conducted at each velocity for each dummy and sufficient time was allowed to elapse between the two experiments. The upper and lower neck forces and moments were determined from load cell measures and its locations with respect to the ends of the neck. Both dummies showed good repeatability for axial and shear forces and bending moments at all changes in velocity inputs. Morphological characteristics in the neck loading responses were similar in all surrogates, although the peak magnitudes of the variables differed. In general, the THOR better mimicked the PMHS response than the Hybrid III dummy, and factors such as neck design and chest compliance were attributed to the observed variations. While both dummies were not designed for use at the two extremes of the tested velocities, results from the present study indicate that, currently the THOR may be the preferred anthropomorphic testing device in crashworthiness research studies and full-scale vehicle tests at all velocities.