The Indo-Asia collision significantly changed the topography and drainage network of rivers around the Tibetan Plateau. Debate continues as to when and how the current drainage system of the Yangtze River was formed. Here we use 40 Ar/ 39 Ar dating of detrital micas (muscovite and biotite) to constrain provenances of the Pliocene sediments from the Jianchuan and Yuanmou basins in SE Tibet. Muscovite and biotite data of the same Pliocene samples from the Jianchuan Basin suggest contrasting distal v. local sources, respectively. Similarly, muscovite data of the Yuanmou Basin suggest a derivation of sediments from the Yalong River, but the characteristics of the Pliocene cobbles (palaeocurrent and subrounded cobbles) suggest that these sediments are locally sourced. Sediment reworking is proposed as an explanation for the different sediment provenance signals in the Jianchuan and Yuanmou basins that have led to the controversy of an either Pleistocene or pre-Miocene age of formation of the current Yangtze. Based on sediment provenance constraints, the evolution of the Jinsha River is reconstructed. The upper Jinsha River lost its connection with the southward flowing Red River upstream from the Jianchuan basin at least before the Pliocene. At the same time a parallel site in the Yuanmou Basin shows that the Yalong River stopped flowing southward into this basin. Detrital mica from early Pleistocene sediments at the Panzhihua site between the Jianchuan and Yuanmou basins is sourced from the current Jinsha and Yalong rivers. These results would suggest that the current upper Yangtze drainage system should have been established before the Pliocene. Supplementary material: Muscovite 40 Ar/ 39 Ar data, biotite 40 Ar/ 39 Ar data and muscovite geochemistry are available at https://doi.org/10.6084/m9.figshare.c.4821573