We studied the neural changes accompanying the learning of a perceptual-motor task involving polyrhythmic bimanual force production. Motor learning was characterized by an increase in stability of performance. To assess after-effects in the corresponding neural network, magnetoencophalographic and electromyographic signals were recorded and analyzed in terms of (event-related) amplitude changes and synchronization patterns. The topology of the network was first identified using a beamformer analysis, which revealed differential effects of activation in cortical areas and cerebellar hemispheres. We found event-related (de-)synchronization of β-activity in bilateral cortical motor areas and α-modulations in the cerebellum. The α-modulation increased after learning and, simultaneously, the bilateral M1 coupling increased around the movement frequency reflecting improved motor timing. Furthermore, the inter-hemispheric γ-synchronization between primary motor areas decreased, which may reflect a reduced attentional demand after learning. © 2008 Elsevier Inc. All rights reserved.