Human movements emerge from complex interactions between the musculoskeletal system, the neural system and the external environment. Their control is realised through a hierarchical neural system composed of the brain, the spinal cord (SC), and remaining nerves connecting to the body limbs. The SC acts as a brain-muscle gateway by integrating descending control signals from higher brain areas, and sensory signals from muscle receptors. The SC also implements its own control mechanisms, including fast sensory reflexes, control of rhythmic locomotion through central pattern generators (CPG), low level control through motor synergies. Nevertheless the full picture of the role of the SC in human motor control remains unclear. The SC is evolutionary old, emerged in the first vertebrates about 500 million years ago, and fully allowed basic locomotion. As new higher neural areas evolved to handle more complex motor control, they had to coexist and interact with the old SC, a support or a constraint? It is suggested that humans present a predominant role of spinal reflexes along with descending modulation over CPG. Movement execution can also be affected by diseases or injuries, such as spinal cord injury and Parkinson's disease. These pathological conditions present various neural and biomechanical motor impairments (muscle spasticity, weakness) in a broad spectrum of severity, and their effect on motor control are hardly separated and identified in clinical studies. Motor rehabilitation strategies that are developed to improve motor control in these conditions such as epidural electrical stimulation (EES) then require a certain personalisation to each patient. Movement execution thus results from interactions between complex systems that are hard to isolate and study. Neurobiomechanical modelling and simulation are then powerful tools to test hypotheses and better understand such complex dynamical mechanisms. Moreover, models can be personalised to study pathological conditions and personalise rehabilitation strategies. This thesis thus aims at investigating the roles of spinal pathways in human motor control using neurobiomechanical modelling tools. Through various collaborations, I studied the SC in various complementary contexts, from healthy control to pathological and rehabilitated control, for both upper and lower limbs. I first studied the role and modulation of spinal pathways for healthy upper limb control in different gravity conditions. I then considered their interaction with cerebellar motor learning for upper limb control. These two studies highlighted the role of the SC in handling perturbations, gravity compensation, and facilitating motor learning. Then I focused on pathological conditions in gait, as it is a movement that is more commonly modelled in previous work, while the cause-effect relationships in pathological conditions are not fully understood. I personalised a model to predict Parkinson patient-specific optimal gait to provide as target to EES rehabilitation strategy. Finally, I personalised an upper limb model to predict quadriplegic patient-specific reachable motor space, and then provide trajectory targets to EES rehabilitation strategy. Overall, neurobiomechanical tools are deemed of great interest to study motor control mechanisms in humans. Moreover, they can assist the design and personalisation of rehabilitation strategies such as EES.