Two experiments were designed to investigate a role of visuo-spatial short-term memory on vocabulary learning of Sign Language as a second language. The subjects having no experiences of Sign Language learning were required to encode Sign Language new words by associating with their meaning (presented in Japanese words) in both experiments. A2×2×3 factorial design was used in experiment 1: the first variable was with or without spatial concurrent task, the second was high- or low-imagery of the Sign Language words (from Matsumi's list, 2000), and the third was trial numbers of paired associate learning. A2×2×3 factorial design was used in experiment 2: the first variable was with or without visuo concurrent task, the second was wide- or narrow- moving of the Sign Language words, and the third was trial numbers of paired associate learning. The main results were as follows: (a) visuo concurrent task interrupted encoding of Sign Language new words, but spatial concurrent task did not, and (b) both imagery and moving wideness of Sign Language were critical factors in words memory. These results suggested that in contrast with phonological short-term memory which is important for speech language, the visuo-spatial short-term memory, especially its visual component, plays an important role in learning Sign Language new words.