This paper engages two issues about trust. First, how to define trust? Does trusting B involve more than merely relying on B? Second, what is the normative structure of trust—a two-place relation or a three-place relation? I defend a new position in the second issue that yields an equally new position in the first. On the standard three-place model, “A trusts B to φ.” On a two-place model, “A trusts B” simpliciter. On my new three-place model, both formulae are less fundamental than “A φs through trust in B.” I call this an Assurance View of trust because I view trusting B as accepting B's implicit invitation to trust—an assurance, in effect, that B is relevantly trustworthy. Trust differs from mere reliance through how it gives a reason grounded in the trusted's worthiness of it. When you trust B, you presume you have a reason to rely on B that is endogenous to your trust, a reason grounded not merely in B's exogenous reliability but in B's responsiveness to your needs. The key insight of the Assurance View emerges when trust is disappointed—B does not φ, though you trusted B to φ—yet not thereby betrayed.