Unlike other modes of loading, the tolerance of the human neck in tension depends heavily on the load bearing capabilities of the muscles of the neck. Because of limitations in animal models, human cadaver, and volunteer studies, computational modeling of the cervical spine is the best way to understand the influence of muscle on whole neck tolerance to tension. Muscle forces are a function of the muscle's geometry, constitutive properties, and state of activation. To generate biofidelic responses for muscle, we obtained accurate three-dimensional muscle geometry for 23 pairs of cervical muscles from a combination of human cadaver dissection and 50 th percentile male human volunteer magnetic resonance imaging and incorporated those muscles into a computational model of the ligamentous spine that has been previously validated against human cadaver studies. To account for multiple origins, insertions, and lines of action, 82 muscle partition pairs, including nonlinear passive and active elastic components, were included in the model. Using optimization, we determined physiologically appropriate levels of muscle activation for each of the 23 muscles simulating relaxed (no pre-impact awareness) and tensed (pre-impact awareness) states. Unlike all prior neck models, both of these states of activation were mechanically stable in an upright neutral anatomic position resisting gravity as an initial condition prior to tensile loading. Tensile forces were then applied to the models at the head center of gravity at 500 N/s. Both states of activation predicted injury above C3, consistent with clinically observed tensile neck injuries. Using these more physiologically reasonable estimates of muscle activation significantly narrowed the range of estimates of 50 th percentile male tolerance to tensile loading. Specifically, while the ligamentous spine fails in tension at an average of 1.8 kN and the total muscle activation predicts neck failure at 4.8 kN, the optimized muscle forces resulted in a whole neck tolerance of 3.1 kN and 3.7 kN for the relaxed and tensed neck, respectively. Moreover, these techniques provide quantitative estimates of load sharing between the neck musculature and ligamentous spine that will be useful in the creation of next generation ATD necks.