<p>Continental collision zones undergo a series of deep processes, including subduction of oceanic lithosphere, slab rollback, slab breakoff, continental subduction, lithospheric thickening, and lithospheric delamination. How these deep processes influence the magma generation and the compositional maturation of continental crust in collision zones remain poorly understood. Based on the consensus that plate motion during the Phanerozoic is primarily driven by the pull force of the subducting oceanic slab, this paper divides the evolution of continental collision zones into four stages, including pre-collision (oceanic subduction), syn-collision, transition, and post-collision, separated, respectively, by initial collision, oceanic slab breakoff, and initial extension occurring at the passive continental margin. Intermediate-felsic magma generated during pre-collision primarily originates from the fractional crystallization of mantle-derived mafic magmas under water-rich conditions, producing a large amount of compositionally complementary hornblende-rich mafic-ultramafic cumulates, whereas intermediate-felsic magma formed during syn- and post-collision is generated primarily by the partial melting of pre-existing hornblende-rich mafic-ultramafic rocks, leaving a large number of eclogitized melting residues. The two-stage process of accumulation and remelting in the collision zone—defined herein as accumelting in this paper—involving the formation of voluminous hornblende-rich cumulates during pre-collision and their remelting during late subduction, syn-collision, and post-collision, has led to the compositional maturation of the continental crust in collision zones. Similar trends in magma compositional changes and deep processes are observed in other collision zones, suggesting that accumelting may be an effective process leading to the generation and compositional maturation of the continental crust. Future research directions should focus on (1) the scarcity of arc magmas and the genesis of intermediate-felsic rocks in continental collision zones, (2) whether eclogitized melting residues or cumulates experienced large-scale delamination or were transferred across the Moho into the seismically-defined upper mantle during syn-collision, (3) whether the extensive I-type K-rich intermediate-felsic rocks (including I-type granitic rocks and extrusive equivalents) in the upper plate of collision zones primarily come from the partial melting of pre-existing K-rich mafic-ultramafic meta-igneous rocks (protoliths include cumulates and basaltic rocks) in the lower crust, and (4) whether the lower crust of magmatic arcs containing hornblende-rich cumulates is an important reservoir for volatiles.</p>