This contribution discusses the methodological status of content analysis. It is argued that content analysis, like interview or observation technique, is an instrument to record social reality of a special type, namely symbolic social reality. Furthermore, since interviewing and observation first transform social reality into symbolic reality (texts), content analysis necessarily always is implied when transforming, in a second step, symbolic reality into data. According to this two-fold process, one can demand two types of reactivity. Conditions, under which reactivity in content analysis necessarily will occur, are specified with regard to semiotic levels. The model to detect reactivity in content analysis, as developed here, is applied to test a content analytic technique which is supposed to be strongly reactive. Finally, suggestions are made to help rule out reactivity in content analysis.