Ijime suicide (or suicide caused by bullying) has been one of the most serious problems in Japanese primary and secondary schools since the late 1970s. The problem has been widely studied in Japan. Generally speaking, the prevailing view is that though bullying has always existed, the situation of children committing suicide because of bullying is getting worse. Against this prevailing view, some scholars in the field of the sociology of education claim that "ijime suicides" existed in the past, and bullying condition may not be getting worse, but what has happened is that people have become more sensitive. In other words, they refer to the social condition itself. However, if we strictly observe the methodological standard of social constructionism, the bracketing of the ontological status of "social problems, " we must conclude that they are making an error of "ontological gerrymandering." We may have a tendency to regard ijime as bullying. However, according to Jeff Coulter, To claim that 'X' is the same action as 'Y' can mean creating a connection between them rather than simply recording a pre-existing relationship. We may misunderstand the property of our description. According to Peter Winch, the human activity we study is carried on according to rules. Therefore, we need to study the rulefollowing discourse of "ijime suicide" carefully. I would claim that "ijime suicide" does not exist until people learn of this term and the discourse. This is not an empirical, but rather a conceptual problem. Consequently, the aim of this paper is to solve this problems and explain "ijime suicide" itself analytically by focusing on the discourse involved, and then to indicate the possibility that a discursive reformation can dissolve "ijime suicide."