Granite saprolites are important hosts of rare earth elements (REE), especially in southern China. The elucidation of REE's migration and fractionation mechanisms during granite weathering allows researchers to thoroughly investigate REE deposits efficiently. This study comprehensively analyzed REE contents and distribution patterns in bulk samples obtained from two granite profiles, exposed to different climatic conditions. In both profiles, REE concentrations increased with depth, indicating downward migration (facilitated by the acidic conditions near the surface) and significant enrichment of REEs at the bottom. The distribution of REEs in the weathering profiles is highly related to mineralogical compositions (such as the role of clay minerals and Fe oxides in hosting REEs). Moreover, mineralogy also plays a crucial role in modulating REE fractionation while others affect secondarily. Substantial amounts of kaolinite and illite preferentially adsorb HREEs and LREEs, respectively; thus, the intensively weathered HZ profile, developing in a warm and humid climate, presents elevating (La/Yb)N toward the bottom. In contrast, considering its distinct mineralogical compositions, the similar patterns of (La/Yb)N of the WMS profile in dry and cold conditions, result from preferential retention of HREEs by chlorite or OM and significant LREEs release via epigenetic denudation. In addition, both profiles exhibited significantly positive Ce and Eu anomalies near the surface, with those in the HZ profile being more severe. The HZ profile, being exposed to warm and humid conditions, was characterized by more pronounced REE migration (61% losses of REEs during weathering) and significant (La/Yb)N (1.24 +/- 1.15, n = 15, 1sd) than the WMS profile (28% losses, 1.06 +/- 30, n = 19, 1sd), which were predominantly controlled by massive clay minerals formed during extreme weathering; correspondingly, the WMS profile, with different mineralogical compositions, experienced less complete chemical weathering, wherein REE behavior would be affected by multiple factors.