Typically, most research on facial expressions has relied on photograph images or “static” information. However, affective facial expressions are dynamic in nature. In this article, the author proposes a morphological approach for elucidating complex spatiotemporal information embedded in dynamic facial expressions. To extract the components of dynamic facial expressions, a combination of the methodologies of geometric morphometrics and a multiway decomposition method, known as PARAFAC, was applied to sequential, three-dimensional landmark coordinate values that were obtained using a facial motion capture system. The result suggests that dynamic facial expressions consist of plural components that differ in spatiotemporal characteristics (i.e., long- and short-term facial deformations).