Dual coding theory proposes that abstract concepts are encoded and stored in memory in the form of symbolic or verbal representations, whereas concrete concepts are dually encoded into memory as both verbal representations and image codes grounded in perceptual experience. This chapter describes the three recent event-related functional magnetic resonance imaging (fMRI) experiments in which word imageability was manipulated during visual lexical decision, semantic decision, and word naming tasks. It attempts to resolve some of the inconsistencies in the imaging literature by using relatively large samples of participants and large numbers of items to ensure reliable activation patterns, by careful matching of concrete and abstract items on all possible nuisance variables, and by matching as closely as possible the task demands for the concrete and abstract conditions. The results were remarkably consistent across studies and provide clear support for some of the basic tenets of the dual-coding model.