It is well established that the central nervous system (CNS) stabilizes the head using reflexive feedback and cocontraction. The major reflexive pathways in the neck are through muscle spindles generating the cervicocollic reflex (CCR) and through the vestibular organ generating the vestibulocollic reflex (VCR). The CNS modulates the contribution of the different pathways and the level of cocontraction to change the system dynamics in an effort to optimally withstand motion disturbances. Predetermined groups of muscles, called synergies, are used to generate stabilizing forces, but it is not clear how the CNS modulates the reflexive pathways and how the muscle groups are chosen. To understand the function of the different reflexive pathways and the responses of the different muscles musculoskeletal models are necessary. To date, a neck model with sufficient detail to simulate vertebral injury and reflexive control of individual muscles does not exist. There are a number of reasons why the behavior of the CNS in neck muscle control should be investigated. Cervical dystonia is a movement disorder characterized by involuntary activity of the neck muscles leading to debilitating abnormal postures and twisting movements. Current evidence points towards changes in the neuronal circuitry in the brain, but the underlying pathology remains for the most part unclear. The understanding of normal and aberrant control of muscles is an important key to finding a solution for this disorder. The current treatment involves injecting the dystonic muscles with botulinum toxin. Although this treatment is generally quite successful, it often remains difficult to distinguish dystonic muscles from healthy compensatory activation. Improvements in the protocols to select aberrant muscles in patients would improve the effectiveness of the treatment and could be an asset when treating more difficult cases. Two aims were set out in this thesis. The first was to establish how the CCR and VCR pathways activate individual muscles to ensure head and neck stabilization and how their gains modulate under different loading conditions. To this end a detailed neuromuscular model was developed containing reflexive neuromuscular feedback and cocontraction. The second aim was to understand the pathology behind cervical dystonia by quantifying aberrance of individual muscle activation in patients, and to develop a protocol to improved muscle selection for possible diagnostic use. In total, five studies were performed to achieve these goals. The first aim of this thesis was achieved by investigating neuromuscular control of healthy subjects using a dynamic experiment and a detailed neuromuscular model. First, we wanted to establish a relationship between the modulation of reflexes and the amplitude and bandwidth of a disturbance to the head and neck (Chapter 2). This was determined by perturbing seated subjects in an anteriorposterior direction with varying amplitudes and frequency content. We found a substantial attenuation of both vestibulocollic and cervicocollic reflexes as the frequency content of the perturbation increased, but only small changes with amplitude. The subjects performed these tasks with the eyes open and closed, and it was observed that with the eyes open subjects were able to further reduce the head motion in space. To investigate how and why the reflex pathways modulate with bandwidth a neuromuscular model was developed. A model including eight vertebrae and the skull and full joint motion was extended with a detailed set of 258 muscle segments. The spinal stiffness, muscle moment arms, instantaneous axes of rotation, and cervical strength were validated in its neutral position (Chapter 3). The importance of a mechanical equilibrium in the neck when estimating model strength was also addressed. The model was able to predict isometric muscle activation patterns of healthy subjects, and the study demonstrated that model strength will be overestimated particularly in flexion and axial rotation the equilibrium over all the neck joints is ignored. Neuromuscular control of the model was then implemented to enable dynamic simulations, which included the vestibulocollic and cervicocollic reflexes and cocontraction. Nonlinear sensory and muscle activation dynamics were implemented along with neural delays for the different feedback pathways. Muscle synergies defined in isometric conditions to generate directional forces under joint equilibrium were used to convert the reflexive vestibulocollic feedback of the semicircular canals and otoliths to individual muscle activations. A similar muscle synergy was used to generate cocontraction over a predefined set of muscles. The cervicocollic reflex was modelled such that each muscle element contained its stretch reflex (muscle length and velocity). The neuromuscular model was subsequently used to investigate the modulation changes with bandwidth observed in Chapter 2 by optimizing reflexive gains and cocontraction of the model to experimental subject data. A possible strategy involved in reflex modulation was also investigated by estimating muscle energy consumption (strategy to minimize effort) and head motion (strategy to stabilize the head). In this study, reflexive gains semicircular and muscle spindle gains were modulated in congruence with our hypothesis, showing gain reductions with increasing bandwidth. However, in contrast to the hypothesis cocontraction was also found to steadily increase with bandwidth. The primary strategy of the CNS appeared to be a suppression of resonant oscillations of the head between 1-3 Hz and, if necessary, between 6-8 Hz. When the perturbation bandwidth increased the reflexes and cocontraction were modulated to dampen the higher frequencies as they were being excited. To investigate aberrant activity of dystonic muscles a standardized protocol was developed by fixing the head of subjects in an isometric device. This ensured a similar position of the head and neck and allowed for controlled task instructions using head force measurements and visual feedback. Individual muscle electromyography (EMG) of 10 cervical dystonia patients and 10 healthy subjects were compared in different loading directions. To addresses a possible worsening of dystonic responses due to an increase in voluntary muscle contractions, activity was evaluated at different contraction levels. Increased levels of mean and minimum (cocontraction) EMG were found in clinically diagnosed dystonic muscles (Chapter 5) and the spectral content of these muscles was shifted (Chapter 6), likely indicating an altered control of the muscles by the central nervous system. The abnormal activity was found during submaximal contractions and in rest, but no differences were observed during maximal contractions. Also, no evidence was found of an exacerbation of dystonic activity related to an increase in muscle contraction when the head was fixed. These results seem to indicate that dystonic muscles are controlled during submaximal contractions much like healthy muscles are activated during maximal contractions. The identified aberrant activity in patient muscles was then evaluated as an identifier of dystonic muscles. The protocols were able to distinguish dystonic activity on a group level, but were not yet sufficiently accurate to be used as a diagnostic tool. Currently, a study is being performed with an improved setup and including patients not previously treated with botulinum toxin to further investigate the applicability of these methods for diagnostics. The contents of this thesis can be summarized with the following main conclusions: I. The central nervous system modulates reflex gains and cocontraction with the frequency content of perturbations with the goal to suppress resonant frequencies of the head while minimizing effort. II. Dystonia patients show increased mean and altered spectral muscle activity during submaximal contractions and rest, but not during maximal contractions. In addition, a higher level of muscle activation does not exacerbate dystonic activity. A neuromuscular model with full joint motion and reflexive control of individual muscles was developed, which is a valuable tool for future investigations into neuromuscular disorders and injury of the neck. The protocols for the identification of aberrant muscle activity in dystonia patients that were established in this thesis have the potential to be used for dystonic muscle selection in the future.