Three experiments were carried out to elucidate the origins of the concreteness (C) effect in a lexical decision task. The first experiment was a replication of the work of Schwanenflugel et al. (1988) and Van Hell and De Groot (1998), who presented the context availability (CA) hypothesis. In this experiment CA seemed to be a dominant factor. Familiarity (FAM) was not incorporated in the ANOVA, but a regression analysis and negative correlation between C and FAM in the groups matched on CA showed that FAM could explain the disappearance of the C effect. Experiment 2 controlled FAM and revealed a C effect, although concrete and abstract words were matched on CA. Experiment 3 controlled C and FAM and revealed a CA effect. The current data emphasize the importance of controlling FAM and CA in examining the C effect in a lexical decision task and support a revised version of the dual-coding theory.