Mirror visual feedback has previously been found to reduce disproportionate interlimb variability and neuromuscular activity in the arm muscles in children with Spastic Hemiparetic Cerebral Palsy (SHCP). The aim of the current study was to determine whether these positive effects are generated by the mirror per se (i.e. the illusory perception of two symmetrically moving limbs, irrespective of which arm generates the mirror visual feedback) or by the visual illusion that the impaired arm has been substituted and appears to move with less jerk and in synchrony with the less-impaired arm (i.e. by mirror visual feedback of the less-impaired arm only). Therefore, we compared the effect of mirror visual feedback from the impaired and the less-impaired upper limb on the bimanual coupling and neuromuscular activity during a bimanual coordination task. Children with SHCP were asked to perform a bimanual symmetrical circular movement in three different visual feedback conditions (i.e. viewing the two arms, viewing only one arm, and viewing one arm and its mirror image), combined with two head orientation conditions (i.e. looking from the impaired and looking from the less-impaired body side). It was found that mirror visual feedback resulted in a reduction in the eccentric activity of the Biceps Brachii Brevis in the impaired limb compared to the condition with actual visual feedback from the two arms. More specifically, this effect was exclusive to mirror visual feedback from the less-impaired arm and absent when mirror visual feedback from the impaired arm was provided. Across conditions, the less-impaired arm was the leading limb, and the nature of this coupling was independent from visual condition or head orientation. Also, mirror visual feedback did not affect the intensity of the mean neuromuscular activity or the muscle activity of the Triceps Brachii Longus. It was concluded that the positive effects of mirror visual feedback in children with SHCP are not just the result of the perception of two symmetrically moving limbs. Instead, in order to induce a decrease in eccentric neuromuscular activity in the impaired limb, mirror visual feedback from the 'unaffected' less-impaired limb is required. © The Author(s) 2011.