Trapezius (Tr) and serratus anterior (SA) are shoulder muscles which help to stabilise the scapula on the chest wall. In a previous study, we described interesting differences in the reflex properties of these two muscles (Alexander et al. 2005). Here we examine the descending control of these muscles using magnetic stimulation of the cortex. With informed consent from 14 subjects, EMG recordings were made from SA, upper Tr and lower Tr muscles using surface electrodes placed bilaterally. First, the optimal sites for stimulation of the motor cortex were examined. Using a grid marked on a cloth attached to the head, magnetic stimuli were applied to the scalp overlying the motor cortex in different positions until the largest amplitude contralateral MEP (cMEP) was evoked for each of the three muscles studied. In these positions, the stimulator output was varied from 20 to 100% of the stimulator output. The latency at 1.2 x threshold for the cMEP was 10.4 ± 1.1, 12.3 ± 1.4 and 12.0 ± 1.1ms (mean ± S.D.) for upper Tr, lower Tr and SA, respectively. With ongoing EMG activity, the threshold of the cMEPs was approx. 40% of the stimulator output. The position of the coil was then systematically moved around the grid. The amplitudes of the cMEPs were then used to construct a surface map. The optimal site was expressed as a centre of gravity (C of G). An average site for the C of G was expressed as the mean anterio-posterior position ± SD, the mean medio-lateral position ± SD. A two-way ANOVA without replication demonstrated that the C of G for each of the muscles recorded could not be separated (SA -0.6 ± 0.8, 3.8 ± 0.5cm, upper Tr -0.5 ± 0.9, 3.7 ± 0.7 cm and lower Tr -0.7 ± 0.7, 3.7 ± 0.6cm). In 10 subjects, this experiment was repeated bilaterally for upper Tr in order to examine the symmetry of this map. In 9 of 10 subjects, the maps were asymmetrical in either the anterior posterior or the medial lateral direction by at least 1.5cm. Finally, in order to study possible ipsilateral connections, the EMG was rectified and averaged. During a weak contraction, ipsilateral responses were evoked in upper Tr at a latency of 19.0 ± 2.7ms (85% occurrence), i.e. nearly 9ms longer than the corresponding contralateral response. In contrast, no ipsilateral MEPs were evoked in SA even at high stimulator outputs and/or with ongoing EMG activity. In conclusion, cMEPs can be evoked in both Tr and SA with latencies consistent with a direct corticospinal connection. While these experiments indicate that the optimal position to evoke MEPs in these muscles are identical, care needs to be taken when evoking MEPs from each side due to inherent asymmetries. Tr and SA differ in their bilateral descending control. Whereas such magnetic stimulation often evoked an MEP in ipsilateral Tr (albeit at a higher threshold and at a long latency), no responses were evoked in ipsilateral SA.