Kimberlites in the eastern United States contain two suites of megacrysts/inclusions that are mineralogically similar but compositionally distinct. One suite (olivine, garnet, diopside, Cr-spinel) has higher Cr and Mg than the other (olivine, garnet, diopside, picroilmenite). Based on detailed petrologic studies of megacrysts from the Fayette County, Pennsylvania, kimberlite, Hunter and Taylor (1984) suggested that these two suites represent the crystallization products of separate magmas that mixed in the low-velocity zone (LVZ) to form kimberlite magma. Major and trace element abundances of individual garnet megacrysts from eastern U.S. kimberlites (i.e., from Kentucky, New York, Pennsylvania, and Tennessee) support the magma-mixing hypothesis but also indicate additional complications. Eclogite garnets have Cr 2 O 3 <0.3 wt.%, CaO >7 wt. %, and chondrite-normalized Lu/Hf <<1. Peridotite garnets have Cr 2 O 3 >2 wt.%, MG# >83, and chondrite-normalized Lu/Hf <1. Garnet megacrysts from Kentucky and Pennsylvania form two groups, one with TiO 2 <0.5 wt.%, and one with TiO 2 >0.5 wt.%. Both groups span a similar range in Cr 2 O 3 (≅ 1.0 to 9.0 wt.% Cr 2 O 3), but the high-Ti garnets may have Cr 2 O 3 as low as 0.1 wt.%. The low-Ti garnets have chondrite-normalized Lu/Hf <1 and are probably derived by the disaggregation of peridotite xenoliths and wall rock. The high-Ti garnet megacrysts have chondrite-normalized Lu/Hf ≥1 and are interpreted here as cognate “phenocrysts” that crystallized in a kimberlite or proto-kimberlite magma. Two suites of high-Ti garnet megacrysts are recognized: a low-Cr to very low-Cr suite (Cr 2 O 3 <4 wt.%) with flat to slightly positive heavy rare-earth element (HREE) slopes, and a high-Cr suite with steeply negative HREE slopes. These suites correspond to the “Cr-poor” and “Cr-rich” suites, respectively, defined by Hunter and Taylor (1984) for the Pennsylvania kimberlite. These data are consistent with the mixing of two magma batches to form kimberlite, as proposed by Hunter and Taylor (1984). Mixing prpobably occurred in the LVZ prior to eruption of the hybrid kimberlite magma.