Predictive conceptual models help us frame research questions, interpret results, and guide clinical practice. Although numerous models have appeared in the assistive technology (AT) literature, none has been shown to predict AT usage. The lack of a valid predictive model indicates the need for development of new approaches to modeling AT outcomes. This article proposes a user-centered conceptual model that predicts AT usage as a function of the perceived relative advantages of AT. Device usage is not modeled as a one-time, all-or-nothing proposition, but as a decision process recurring over time. The influence of parallel interventions working concurrently with, or as an alternative to, AT is a central consideration that ultimately drives AT usage. Usage is shown as a proximal influence on AT impact, and AT impact is shown to be a predictor of future use. Research is cited supporting various elements of the new model.