This article consists of three parts. In the first part, early empirical research on music performance is reviewed—with special emphasis on the contributions by C. E. Seashore and his co-workers at Iowa University in the 1930s. The second part presents a model for interplay between analysis and synthesis in studies of music performance and its relationship to listeners' experience of the music. The model means that music performance is analyzed with regard to various physical properties, and their relationships to listeners' experience are investigated by means of synthesized sound sequences that are systematically varied in different aspects. In the third part, this idea is illustrated by examples from an extensive research project on musical rhythm. It is shown that performance of musical rhythm is characterized by various systematic variations regarding the durations of the sound events relative to strict mechanical regularity and that these variations may be related to various aspects of the experienced rhythm.