This paper explores different options for establishing the matrix language in Russian-Estonian code-switching. First, the notions of matrix and embedded language are introduced, followed by a short overview of the Russian-Estonian bilingual situation in Estonia. The Matrix Language Frame model (Myers-Scotton 1993, 1997, 2002) is introduced in the third part of the paper, where the corporal data are described as well. A possible application of the Matrix Language Frame model to the Russian-Estonian code-switching data is presented in part four. It is argued that in most cases the matrix language is clearly definable. However, there are instances which can be best described as “congruent lexicalisation” (Muysken 2000).