Human respiratory motoneurones in the spinal cord receive drives that originate from multiple supraspinal sites, depending on the ‘task’, i.e. from the medulla for quiet breathing and the motor cortex for voluntary tasks. The integrated motoneurone output can be assessed by intramuscular single motor unit recordings. During quiet breathing, there is differential output from motoneurone pools of the human obligatory inspiratory muscles (Saboisky et al., 2007). The timing, relative to inspiratory flow, and the magnitude of motor unit activity is non-uniform. Specifically, for the parasternal intercostal muscles of the first-to-fifth interspaces, motor units in the rostral interspaces are active early in inspiration and discharge at a high rate compared to caudal interspaces (Gandevia et al., 2006). Remarkably, the degree of inspiratory motor unit output of the parasternal intercostal muscles parallels the spatial distribution of inspiratory mechanical advantage of these muscles across interspaces. Consequently, a principle of “neuromechanical matching” between the neural drive and mechanics of portions of inspiratory muscles may govern recruitment of spinal respiratory motoneurones when driven from the pontomedullary region during quiet breathing. In targeted voluntary breaths, when drive originates from the motor cortex, the rostro-caudal pattern of motor unit output from the first-to-fifth parasternal intercostal is maintained (Hudson et al., 2011b). The parasternal intercostal muscles are also active in a different voluntary task of ipsilateral trunk rotation. The voluntary postural and inspiratory drives depolarise the same motoneurones and a concurrent postural task changes the inspiratory output of the parasternal intercostal motoneurones in a direction-dependent manner (Hudson et al., 2010). However, in a similar rotation task, the costal diaphragm is not active and the inspiratory output of the phrenic motor units is not altered (Hudson et al., 2011a). Taken together, these observations have important implications for the integration of voluntary and inspiratory drives to respiratory motoneurones at different level of the spinal cord. We propose that premotoneuronal networks, perhaps in the spinal cord, sculpt descending drive to respiratory motoneurones from multiple sources (Hudson et al., 2011c). New data support this proposal. The recruitment behaviour of parasternal intercostal motor units of different interspaces was assessed in distinct inspiratory and voluntary rotation tasks. As expected, there was differential inspiratory activity in motor units from the 2nd and 4th interspaces with earlier onset of activity in the rostral interspace, relative to the onset of inspiratory flow. However, for the same motor units, there was no difference in the rotation torque at which the units were recruited during ramped ‘isometric’ rotations. With voluntary drive for the rotation task, there is divergence from the differential recruitment observed during inspiration. This suggests that parasternal intercostal motoneurone output at different spinal levels can change depending on task and that the output of respiratory motoneurones may be related to the precise mechanical advantage of the muscles for that task. A spinal mechanism that integrates and distributes the drive to different human inspiratory muscles would determine the differential pattern of activation across inspiratory muscles, preserve the neural and mechanical coupling when voluntary breaths are taken and allow for different patterns of activation in non-respiratory contractions. Studies in dogs reveal that high-frequency stimulation at the T2 spinal level mimics the physiological activation of the parasternal intercostal muscles (and other inspiratory muscles) during breathing (e.g. DiMarco & Kowalski, 2015). The differential activity between regions of the parasternal intercostal muscles is preserved in these animals, including differences between the medial and lateral portions of a muscle within an intercostal space. As the spinal cord was transected at the C1 level during stimulation, these data corroborate our findings in humans that spinal cord contains circuits that can distribute drive to spinal respiratory motoneurone pools.