In this paper, we investigate the relation between head movement and the synthesis-periphrasis distinction in the verbal domain. We use the term "synthesis" to refer to verbal expressions in which the lexical verb bears all the verbal inflection in a clause. In contrast, a periphrastic verbal expression additionally contains an auxiliary verb ("be" or "have"), and verbal inflection is distributed between the lexical verb and the auxiliary. We argue for a one-way correlation between head movement and periphrasis, first observed in Pietraszko 2017, which we refer to as the "Synthesis-Movement Generalization": If the lexical verb moves to T, it must form a synthetic expression with T, that is, periphrasis is impossible. Importantly, the inverse correlation doesn't hold, as synthesis is compatible with, but does not entail, head movement of the lexical verb to T. We argue that existing theories of the synthesis/periphrasis distinction are either too weak and predict no correlation at all, or too strong and predict a two-way correlation. We further argue that the Synthesis-Movement Generalization can be explained by the hypothesis that both head movement and periphrasis are related to selection. More specifically, we propose that head movement is parasitic on a selectional relation (following Svenonius 1994, Matushansky 2006, and Preminger 2019) and that periphrasis is Merge of an auxiliary verb triggered by a selectional feature of T (Déchaine 1995, Cowper 2010, Pietraszko 2017). This analysis derives the complementarity of head movement of the lexical verb to T and the presence of an auxiliary from the fact that T’s selectional feature can be satisfied by the lexical verb or an auxiliary, but not both. If T’s selectional requirement is satisfied by an auxiliary, it’s not satisfied by the lexical verb, so the lexical verb can’t undergo head movement to T.