The purpose of this study is to configure the landscape of empirical mathematics educational research on problem-solving in teacher education, and thereby disentangle how mathematical problem-solving is understood and used. The method consists of a configurative review of empirical mathematics education research on problem-solving in teacher education. A two-dimensional model is presented to illustrate how different aspects of problem-solving in teacher education are connected to and complement each other. Using the model, the configuration results in the proposition of four major categories of research on problem-solving in teacher education. The result indicates an almost equal distribution of research which views problem-solving as an aim for mathematics education versus research which views problem-solving as a means for learning mathematics. However, within the former, roughly three quarters of the articles focus on content knowledge, and only a quarter on pedagogical content knowledge. Implications for teacher education and future research are discussed.