Mental imagery improves paired-associate (PA) learning relative to overt rehearsal. The effect might be due to increased reliability of stimulus encoding or to increased relational association produced by imagery. These hypotheses expect different outcomes when imagery and rote-rehearsal Ss are compared on memory tests of stimuls recognition and on recall of the response term conditional upon stimulus recognition. Ss learned Pas using one of three methods—rose repetition, interactive imagery, or separation imagey. Associative recall was highest for inteactive-imagery. Ss and lower and equal for rote- and sepration-imagery Ss. No differences in stimulus recognition appeared. Such evidence supports the relational-organizing interpretation of the PA effect of imagery in opposition to the stimulus-distinctiveness or reliable-encoding explanations.