Audiovisual perception and imitation are essential for musical learning and skill acquisition. We compared professional pianists to musically naive controls with fMRI while observing piano playing finger-hand movements and serial finger-thumb opposition movements both with and without synchronous piano sound. Pianists showed stronger activations within a fronto-parieto-temporal network while observing piano playing compared to controls and contrasted to perception of serial finger-thumb opposition movements. Observation of silent piano playing additionally recruited auditory areas in pianists. Perception of piano sounds coupled with serial finger-thumb opposition movements evoked increased activation within the sensorimotor network. This indicates specialization of multimodal auditory-sensorimotor systems within a fronto-parieto-temporal network by professional musical training. Musical ''language,'' which is acquired by observation and imitation, seems to be tightly coupled to this network in accord with an observation-execution system linking visual and auditory perception to motor performance.