The present study applied the false memory experimental paradigm to the functional independence-interdependence issue of bilinguals. Seventy-four Japanese university students who had learned English for at least seven years participated in a list-learning experiment. Most of them were considered as Japanese-dominant bilinguals. Twelve 15-word-list (180 words) were successively presented. Each list was constructed so that a non-presented word (CNW) would be falsely recognized. Six lists were presented in Japanese and the other six lists were presented in English. Half of the recognition test words were presented in English and the rest in Japanese. These factors were orthogonal. The results showed that, although hit rates were higher when the list language and the test language matched, false-alarm rates of CNWs were higher when the test language was Japanese. These supported the unbalanced, independent storage model, which was proposed in the present study.