Adopting the concept that "decisions can be seen as argument-driven actions," the purpose of this study was to apply the argumentation structure raised by Stephen Toulmin (1958) to capture two science teachers' instructional decision-making mechanisms in their teaching practices. The two case teachers were chosen because of their close estimations of their students' achievement outcomes and both were in their transition stages from competent teachers to proficient ones. A science teaching observation coding schedule (STOCS) was designed to be used in classroom observations, and all these quantitative data were collected to be converted into four issues for the two teachers to justify their ways of teaching. A semi-structure interview was conducted to analyze why and how these two case teachers made their instructional decisions. The results indicate that although they knew that their students could not fully understand what they taught in the class, their teaching strategies were still teacher-dominated modes. What really influenced their instructional decisions included external context factors (e.g., examination pressure, subject contents, limited time and classroom management) and internal experiential factors (e.g., personal educated experiences, beliefs and understandings of constructivism and inquiry). Finally, the authors suggest that science teachers' instructional decision making mechanisms can be appropriately represented by case teachers' argumentations structure.