A statistically average or prototypical member of a category (e.g., human faces) is often perceived as more attractive than less typical category members. Two experiments explored to what extent this may hold for music performance, an artistic domain in which individuality (i.e., deviation from prototypicality) is highly valued. In Experiment 1, graduate student pianists judged 11 student performances of Schumann's "Traumerei," one of which was created by forming the mathematical average of the other 10. The average performance was rated second highest in quality, even though it was judged second lowest in individuality. In Experiment 2, pianists judged 30 performances of the beginning of Chopin's Etude in E major, synthesized so as to vary only in expressive timing and tempo. The timing patterns were derived from expert pianists' recordings and from casual student performances, and they included separate and combined averages. All three averages received high quality ratings, and the expert average was rated highest of all 30 performances. There was a negative linear correlation between rated quality and individuality. Paradoxically, therefore, the students' expressive timing patterns were preferred over the experts'. Possible explanations of this finding are discussed, such as interactions between timing and other performance parameters and conditions under which conventionality tends to be favored over individuality.