Locomotion of animals is controlled by central pattern generators (CPGs) located in the spinal cord. Afferent feedback allows animals to adapt locomotion to the environment. Responses of neural activities to afferent feedback stimulation have been previously examined during fictive locomotion in immobilized cats. However, the locomotion of behaving animals is governed by dynamic interactions between central neuronal circuits, afferent feedback, the musculoskeletal system, and the environment. To study the specific role of afferent feedback in these interactions, we created a neuromusculoskeletal model that included two half-center CPG models controlling a musculoskeletal model of cat hindlimbs. The model was used to investigate the role of afferent feedback in adaptive locomotion, specifically when a foot stepped into a hole or when the model walked on a split-belt treadmill. Particularly, we eliminated various sensory afferents and conducted dynamic simulations of walking. As a result, we found that specific sensory afferents, depending on the situation, contribute to adaptive interlimb coordination to prevent stumbling.