The aim of this paper was to investigate how the faces of persons with different expressions are represented in memory. The subjects encoded three photographed faces for each of 24 test persons. The main conditions were four combinations of face repetitions (same and different expressions) and their presentation intervals (massed and spaced). This was followed by an unexpected yes-no recognition test in which identical pictures of the target faces or the same person's expression-changed faces were randomly presented with distractor faces. The same-picture recognition test showed that the same-expression repetition produced higher recognition than the different-expression repetition. The expression-changed recognition test showed that under the different-expression repetition, the massed presentation produced higher recognition than the spaced presentation, but under the same-expression repetition, there was no difference between the massed and spaced presentations. These findings suggest that the storing of more expressions facilitates face representation.