Day and Bellezza (1983) rejected a dual coding imagery explanation for the superior recall of concrete words because unrelated concrete pairs were rated lower in composite imagery but were still remembered better than related abstract pairs. We show that dual coding theory explains their results and our new findings using the same paradigm. In Experiment 1, 120 subjects rated imagery or relatedness for 108 pairs that varied in concreteness, pair relatedness, and associative strength. Incidental cued recall followed. Relatedness and strength affected imagery ratings, as did concreteness, and very low relatedness partly accounted for the low composite imagery ratings for unrelated concrete pairs. Concreteness and relatedness also affected recall, and superior recall for unrelated concrete pairs occurred consistently under imagery but not under relatedness instructions. In Experiment 2, 40 subjects rated imagery value and recalled 24 pairs. Subsequent questioning indicated that composite images were retrieved better given stimuli from unrelated concrete than from related abstract pairs. These findings and Day and Bellezza's original results are explained in terms of (1) imaginal and verbal associative processes, which jointly influence composite imagery ratings and recall, and (2) the critical role of stimulus concreteness during image retrieval and recall (i.e., the conceptual peg hypothesis).Dual coding theory (Paivio, 1971(Paivio, , 1986 explains cued recall in terms of the joint activity of independent verbal and nonverbal (imaginal) cognitive systems. Recall depends partly on the capacity of the nonverbal system to generate composite images to pairs during study trials and to redintegrate those images to cue words during test trials (d. the conceptual peg hypothesis, Paivio, 1969). The ease with which such images are generated and redintegrated depends partly on verbal to imaginal referential connections, which are more available for concrete than for abstract words, and partly on associative connections between units within the verbal and imaginal systems. This theoretical account has been supported by positive effects on associative memory of noun concreteness and imagery instructions (see Paivio, 1969), and of associative relatedness (e.g., Kusyszyn & Paivio, 1966). Recently, however, Day and Bellezza (1983) obtained results that they took as evidence against the theory's imagery explanation for concreteness effects in cued recall. Day and Bellezza (1983) had subjects form composite images to pairs of words that varied orthogonally in concreteness and pair relatedness. Subjects rated the vividness of their interactive images and were later tested for cued recall. The critical findings were that subjects rated images to related abstract pairs (e.g., democracy-liberty) as more vivid than images to unrelated concrete pairs (e.g., cheese-fur), but they nonetheless recalled more unrelated concrete than related abstract words. According to Day and Bellezza, such findings contradict dual coding theory and other ...