We performed experiments at 3.0 GPa and 1530-1565 degrees C to investigate the effects of crystal composition on trace element partitioning between garnet and anhydrous silicate melt. Bulk compositions along the pyrope (Py: Mg 3 Al 2 Si 3 O 12 )-grossular (Gr: Ca 3 Al 2 Si 3 O 12 ) join, doped with a suite of trace elements (Li, B, K, Sc, Ti, Sr, Y, Zr, Nb, Cd, In, REE, Hf, Ta, Th, and U) produced homogeneous garnets, ranging in composition from Py 84 Gr 16 to Py 9 Gr 91 , in equilibrium with melt. Trace element partition coefficients (D-values), measured by SIMS, depend greatly on the Mg/(Mg+Ca) of garnet. For example, from Py 84 to Py 9 , D La increases from 0.004 to 0.2, whereas D U increases from 0.029 to 0.42. These variations can be explained by the lattice strain model of Blundy and Wood (1994), which describes trace element partitioning of an element i in terms of the ionic radius of i (r i ), the size of the lattice site on which i partitions (r o ), the Young's modulus of the site (E), and the (theoretical) partition coefficient D o for an ion of radius r o . For trivalent cations substituting in the garnet X-site (Y, REE, Sc, and In), apparent values of r o fitted to our data vary systematically from 0.935+ or -0.004 Aa (Py 84 ) to 0.99+ or -0.01 Aa (Py 9 ), a trend consistent with variations in the size of the X-site. Values of D o show an increase from Py 9 (D o = 2.8+ or -0.1) to Py 84 (4.8+ or -0.1) and Young's modulus E varies from 257+ or -20 GPa for Py 60 to 590+ or -40 GPa for Py 84 . These results allow a quantitative assessment of the influence of crystal chemistry on garnet-melt D-values, thereby forming the basis for a predictive model similar to that recently developed for clinopyroxene-melt partitioning by Wood and Blundy (1997). Our new data emphasize the importance of taking into account crystal composition when modeling trace element behavior in natural systems.