In the external intercostal muscles, electromyographic (EMG) activity follows a rostrocaudal gradient and the muscle is the most activated in the upper part of the chest. In the present studies, we investigated factors that could be involved in this phenomenon, and in particular the contribution of segmental reflexes. To do so, we evaluated in the rabbit the density of muscle spindles using a modified Duron's method. This density, expressed in number of spindles / unit cross-sectional area, was similar in the upper interspaces (2^d to 5^th) and averaged 0.50 spindle / mm². However, this density became statistically different in the sixth and the seventh interspaces and averaged 0.29 0.05 and 0.26 0.03 spindle / mm², respectively. The sensitivity to segmental reflexes was then assessed by altering the displacement of the lower rib during inspiration. The change in EMG activity was quantified and the slope of the exponential regression between EMG activity and rib displacement averaged -0.19 0.03 and -0.25 0.07 for the 3^th and 5^rd interspaces, respectively (ns). In parasternal muscle, the EMG activity was less correlated to the axial rib displacement (slope = -0.03 0.01; P < 0.05 vs external intercostal muscles). In absence of activity, the sensitivity to segmental reflexes was not tested in the 7^th interspace. Finally, the gradient of EMG activity after dorsal rhizotomy was studied. Dorsal root was sectioned at the third, fourth and fifth thoracic levels. The heterogeneity of activity remained present in the medio-lateral direction with the posterior bundle always more active than the middle and medial bundles. The rostro-caudal gradient remained also present and EMG activity was consistently higher in the 3^rd interspace than in the 4^th and the 5^th interspaces. So, heterogeneity in the density of muscle spindles does not explain the gradients of activity. Indeed, the EMG sensitivity to muscle lengthening is similar in the different muscles and heterogeneity of activity persisted after dorsal rhizotomy. These gradients could, therefore, probably result from differences in the central respiratory drive or in its projection on the pool of external intercostal motoneurones.