The authors developed two versions of a therapeutic presence measure, based on an earlier model of presence (Geller & Greenberg, 2002)-Therapeutic Presence Inventory-therapist (TPI-T) and client (TPI-C) versions-to measure in-session therapeutic presence. They explored their reliability and validity in two studies. In the first, items generated from the previously established model were subjected to analyses and expert ratings. In the second study, therapists and clients rated therapists' presence postsession. Therapists also completed the Relationship Inventory, and clients assessed two additional factors: session outcome, using the Client Task Specific Measure-Revised, and therapeutic alliance, using the Working Alliance Inventory. Findings revealed that both versions of the TPI had good reliability and construct validity. However, TPI-T had low predictive validity and the TPI-C showed good predictive validity. In particular, clients reported positive therapeutic alliance and change following sessions when they felt their therapist was present with them.