Three experiments were conducted to examine the recognition process of the jukugo (two-kanji-compound word) from the following three points of view: (1) how jukugos are retrieved from memory (Exp. 1), (2) how jukugos are activated in the retrieval process (Exp. 2), and (3) how the representation of jukugos is structured in memory (Exp. 3). The experiments were carried out using priming paradigms in which subjects performed a lexical decision task. The results of Exp. 1 and Exp. 2 showed that the first kanji was used as retrieval cue for each jukugo and the activation of the first kanji facilitated the identification of jukugo in terms of meaning. In Exp. 3, it was suggested that the lexicon of jukugo was formed in such a structure that several jukugos containing in common the same first kanji are tied together, centering on the first kanji.