This study has two purposes. The first is to determine whether subordinates employ alternative combinations of emotion regulation strategies toward their supervisors beyond merely using surface and deep labor from the person-centered perspective. The second purpose is to understand why such acts of emotion regulation occur in interactions between employers and employees in the typical workplace. Utilizing latent profile analysis on data from 232 office employees in Beijing, China, collected using a two-stage sampling technique, four distinct supervisor-directed emotional labor profiles (i.e., deep actors, non-actors, moderators, and regulators) are identified. We find that these profiles are differentiated by several factors (i.e., individual identity, relational identity, and LMX orientations). Moreover, our findings suggest that employees exhibiting high levels of relational identity are more predisposed to act as deep actors, whereas individuals with high levels of individual identity are prone to being regulators as opposed to becoming deep actors, non-actors, or moderators. In addition, our results also suggest that LMX orientations have moderating effects on the relationships between self-identities and supervisor-directed emotional labor strategies. Overall, the results of this study expand the potential dimensionality of supervisor-directed emotion regulation strategies (e.g., regulating and non-acting) and bridge a gap in our understanding of the factors impacting supervisor-directed emotional labor.