This paper shows in some detail that the hypothesis briefly discussed in Busquets and Denis (Cahiers de grammaire 26: 55–74, 2001) that French displays both PF-deletion and pro-form ellipsis of sentential complements with different classes of verbs is indeed correct. It subsequently focuses on the PF-deletion type (i.e., modal ellipsis) and argues that like English VP ellipsis, this type of ellipsis should be syntactically derived by way of movement and that French elided TPs stand in a topic position. The syntactic chain thus created by Move is composed of two links: the first merge copy, which is always silent, as is the case with most types of movement, and the remerge copy, which also remains silent, being recoverable from the discourse, but sometimes allows phrasal subparts to undergo further A′-movement to Spec, FocP, a position where they are fully spelled-out at PF. The absence of intervention as well as freezing effects associated with the latter movement are then explained using the rescue by PF deletion account developed in Boškovic (Linguistic Inquiry 42: 1–44, 2011). This novel theory of French modal ellipsis is shown to correctly predict the existence of what Busquets and Denis (Cahiers de grammaire 26: 55–74, 2001:13) named French Pseudo-Gapping, as well as the much more restricted distribution of modal ellipsis in untensed contexts. It also provides a partial answer to why French does not have VP ellipsis and why English does not have modal ellipsis.