For subduction-related porphyry Cu deposits, it is generally agreed that the ore-forming magmas fractionate from mafic magmas produced by slab fluid±melt-fluxed melting of the asthenospheric mantle wedge. This model, however, is not able to explain post-subduction porphyry Cu deposits that formed after the termination of slab subduction. Popular models suggest that post-subduction porphyry Cu magmas are partial melts of former lower crustal sulfide-rich arc cumulates with or without minor contribution from potassic mafic magmas. Here we re-appraise this model via a combined approach of natural sample studies and piston-cylinder experiments. The Sanjiang region of SW China is a typical post-subduction porphyry Cu metallogenic belt that formed during the India-Asia continental collision. Based on a detailed petrographic study, we investigated whole-rock elemental and Nd-Sr-Pb isotopic compositions of variously evolved magmatic rocks and garnet-bearing xenoliths throughout the southern Sanjiang region. The results demonstrate that the continental collision-related porphyry Cu magmas formed solely by fractionation of potassic mafic magmas, and that the lower crustal partial melting model built upon wrong radiogenic isotopic compositions of the investigated garnet-bearing xenoliths. The potassic mafic magma fractionation model is further supported by crystallization experiments using a basaltic trachyandesite sample as starting material. The experimental run products reproduce not only the mineral crystallization sequence of magmatic rocks in the southern Sanjiang region but also the liquid line of descent defined by the whole-rock data. Hence, we suggest that the role of pre-existing lower crustal arc cumulates in forming post-subduction porphyry Cu magmas has been greatly overestimated.