Ga/No conversion (GNC), a case alternation between nominative and genitive in Japanese, has been argued in almost every grammatical paradigm proposed to date (Harada 1971, Shibatani 1976, Miyagawa 1993, Ura 1993, Ochi 2001, Hiraiwa 2005). Overviewing the previous syntactic approaches, this talk shows the basic property of GNC. In addition, I clarify its probabilistic distribution from a statistical point of view, examining its relative preference associated with factors such as adjacency and predicate type, and provide evidence of ongoing change in GNC, which was first pointed out by Harada (1971).