The subjects listened to a list of 21 words read aloud and memorized eight words among them indicated by the cue. In VC (voice-cue) condition, where the change of voice between male and female cued the difference between the to-be-memorized and not-to-be-memorized items, the percentage of correct recall was higher and the number of intrusion-errors was fewer than in NYC (non-voice-cue) condition, where the cue was the sound of chime given immediately before the to-be-memorized item. The results suggest that the physical characteristics of cue facilitate the selective memorization, but do not necessarily support the early-selection theory of attention. Next, in order to confirm Watanabe's (1976) assertion that the transformation of to-be-memorized items into long-term memory and the exclusion of not-to-be-memorized items take place in parallel, the subjects were required to rehearse aloud every word as it was presented. However, it was found that the method of voiced rehearsal was inadequate to test the assertion.