As an orchestra has many kinds of instruments, its score consists of many parts (staves of instruments) and needs large amount of space to express even a small piece of music. This makes the musical structure of a score difficult to understand. Musicians must be proficient in score reading before they can understand the structure of a score. To solve this problem, this paper proposes a method to illuminate orchestral scores automatically to help players for easy score reading by coloring musical staves. This method, called ScoreIlluminator, introduces four features to calculate the similarity between parts. These features represent rhythm, sonic richness, melody and harmony. Each indivisual part is clustered by the k-means algorithm using these four features. Our empirical results show that our method succeeds in improving the readability of a score by illuminating similar parts with colors.