It is generally believed that the Tibetan Plateau is the result of crustal thickening in response to the collision of the Asian and Indian plates. However, the specific timing and uplift mechanism remain controversial. The widespread occurrence of Cenozoic lavas in the northern Qiangtang terrane provides a unique opportunity to constrain the dynamic processes that resulted in uplift of the northern Tibetan Plateau. Eocene lavas from the northern Qiangtang terrane display adakitic geochemical characteristics, such as high SiO2 and Al2O3 contents, low Y and Yb contents, positive Sr anomalies, and high Sr/Y and La/Yb ratios, in combination with high Mg# (43–69) and negative anomalies for Nb and Ta, which suggest a garnet + rutile-in and plagioclase-free source residue. The same samples also have high K2O and Th contents, high Th/Ce ratios, and low Nb/U, Ce/Pb, Ti/Eu, and Nd/Sm ratios, as well as high 87Sr/86Sr(i) (0.7062–0.7075) and low εNd(t) (–6.3 to –2.9), which show a clear continental crust affinity. These high-Mg# adakitic rocks, combined with other characteristics of Tibetan Cenozoic lavas, indicate that they were derived from partial melting of delaminated lower continental crust, which subsequently reacted with surrounding mantle peridotites during ascent to crustal depths. The Eocene high-Mg# adakitic rocks (46–38 Ma), north-south–trending shoshonitic dikes (47–38 Ma), and contemporaneous mantle-derived Mg-rich potassic and shoshonitic lavas indicate that the thickness of the crust was at least 50 km before ca. 46 Ma, at which time rapid uplift and extension occurred, most likely caused by small-scale delamination of the lithospheric mantle at 46–38 Ma (Eocene) in central Tibet.