The peroxisome proliferator-activated receptor ␥ (PPAR␥) is a target for treatment of type II diabetes and other conditions. PPAR␥ full agonists, such as thiazolidinediones (TZDs), are effective insulin sensitizers and anti-inflammatory agents, but their use is limited by adverse side effects. Luteolin is a flavonoid with anti-inflammatory actions that binds PPAR␥ but, unlike TZDs, does not promote adipocyte differentiation. However, previous reports suggested variously that luteolin is a PPAR␥ agonist or an antagonist. We show that luteolin exhibits weak partial agonist/ antagonist activity in transfections, inhibits several PPAR␥ target genes in 3T3-L1 cells (LPL, ORL1, and CEBP␣) and PPAR␥-dependent adipogenesis, but activates GLUT4 to a similar degree as rosiglitazone, implying gene-specific partial agonism. The crystal structure of the PPAR␥ ligand-binding domain (LBD) reveals that luteolin occupies a buried ligand-binding pocket (LBP) but binds an inactive PPAR␥ LBD conformer and occupies a space near the -sheet region far from the activation helix (H12), consistent with partial agonist/antagonist actions. A single myristic acid molecule simultaneously binds the LBP, suggesting that luteolin may cooperate with other ligands to bind PPAR␥, and molecular dynamics simulations show that luteolin and myristic acid cooperate to stabilize the ⍀-loop among H2Ј, H3, and the -sheet region. It is noteworthy that luteolin strongly suppresses hypertonicity-induced release of the pro-inflammatory interleukin-8 from human corneal epithelial cells and reverses reductions in transepithelial electrical resistance. This effect is PPAR␥-dependent. We propose that activities of luteolin are related to its singular binding mode, that anti-inflammatory activity does not require H12 stabilization, and that our structure can be useful in developing safe selective PPAR␥ modulators.