Three experiments were conducted to determine the effectiveness of verbal and imaginal rehearsal strategies in children's discrimination learning. With verbal materials, imaging the referent of the correct item was more facilitative than vocalizing the correct item, as long as the former strategy was defined in a manner conducive to effective imagery generation. In fact, when children engaged in such a strategy, their verbal discrimination performance was virtually error-free. No difference between the two strategies was found in pictorial discrimination learning, both resulting in performance very near ceiling. A possible interpretation of the results is provided. (Author)