The present study had the purpose of investigating empirically the structure and relationships among mathematical imagination, mathematical knowledge and mathematical mindset. The three factors are constituent parts of the Innovation Engine model by Seelig (inGenius: A Crash course on creativity, HarperOne-2012) and can influence creative thinking. The participants were two hundred seventeen sixth grade students from three urban and eight rural primary schools. The data were collected by administering a mathematical imagination test, a test measuring mathematical knowledge and a questionnaire capturing their mathematical mindset. Partial least squares structural equation modeling was applied through Smart PLS in order to examine empirically the proposed path model describing the relationships among the three factors contributing to mathematical creative thinking, namely, imagination, knowledge and mindset. The data analysis yielded that the proposed model of the paper fulfilled all evaluation criteria of partial least squares structural equation modeling. In short, students' mathematical knowledge could be explained directly and moderately by their mathematical mindset. In addition, mathematical imagination could be explained directly and to a large extent by mathematical knowledge and indirectly by mathematical mindset. Finally, the main implications for mathematics teaching are discussed and potential research directions are suggested.