Multiple petrogenesis models have been proposed for granitic rocks and the mafic microgranular enclaves (MMEs) hosted in them. The mantle and crust-derived magma mixing/mingling process was widely suggested for the formation of MMEs or the host granitoids. However, whether the mixing between magmas derived from different crustal sources may also form microgranular enclaves (MEs, not restricted to mafic) or the host granites remain unclear. Here, we report petrological, geochemical, and zircon in situ Hf-O isotopic data for MEs and their host granites in the northern Nyima area, southern Qiangtang terrane (SQT), central Tibet. The host granitic rocks consist of amphibole-bearing biotite granites and syenogranite, minor granodiorite and aplite veins, and MEs comprising diorite and granodiorite. Host rocks and MEs are contemporaneous and were generated in the Early Cretaceous (∼120 Ma). Host granitoid samples are high-K calc-alkaline rocks and have high SiO2 (66–78 wt%) and low MgO (<1.14 wt%). They have enriched whole-rock εNd(t) (−1.82 to −3.06), and variable zircon εHf(t) (+0.64 to +5.12) and δ18O (7.35 to 9.61‰) values. The host granodiorites contain some mafic mineral-rich clots and lower silica contents than the biotite granites. They exhibit distinctive geochemical patterns (such as high K2O > 5 wt%, high Zr >500 ppm with Zr/Hf ratios >45, and high Ba >400 ppm), implying biotite and zircon concentrations. Combining this information with embayed feldspars, curved biotites, and closely spaced feldspar grains, we suggest that liquid loss resulted from crystal compaction or filter pressing. Both the fine-grained syenogranites and aplite veins exhibit high SiO2 and Rb/Sr ratios in addition to low Zr/Hf ratios and Zr and Ba concentrations with remarkably negative Eu anomalies, indicating that they are highly differentiated melts extracted from felsic crystal mush. Such a crystal mush is possibly represented by the biotite granites. Gradual contact relationships between variable granitic phases indicate that melt extraction and crystal concentration were coeval. Residual melt extraction from the crystal mush was facilitated by mush remobilization resulting from magma replenishment. We suggest that for the host rocks, the biotite granites were mainly derived by partial melting of ancient crust, and the granodiorites most likely resulted from the magma mixing/mingling that might be accompanied with the loss of melts, the fine-grained syenogranites and aplites represent highly evolved residual melts. MEs are geochemically characterized by intermediate to felsic compositions (SiO2 = 56.3–67.8 wt%), with low MgO (<4 wt%) and Mg# (< 53) and low mantle compatible trace element (Cr and Ni) contents. They exhibit enriched whole-rock Nd (εNd(t) = −0.36 to −2.71), depleted zircon εHf(t) (+0.51 to +5.68) and high δ18O values (7.45 to 9.57‰), except for one sample with comparable whole-rock εNd(t) (−1.32), relatively high zircon εHf(t) (+5.05 to +7.11) and δ18O (6.09 to 7.10‰) values. MEs exhibit evidence for quenched magma crystallization, chemical exchange, mineral disequilibrium and mass transfer, indicating inefficient magma mixing/mingling. Mafic minerals from MEs were in equilibrium with evolved melts rather than mantle-derived primitive melts. Given the substantial absence of contemporaneous basaltic rocks from the SQT and the relatively high SiO2 and low MgO contents and crust-like zircon stable oxygen isotope of MEs, we propose that they most likely resulted from the mixing/mingling of diverse crust-derived magmas. Therefore, our study provides important evidence for crust-derived magma mixing and melt extraction from mush.