以往對於數學解題的研究，偏重於解題結果的探討，但此種研究方法，並無法深入探討學生解題歷程所產生的解題行為。針對這個評量的缺失，放聲思考法(thinking aloud)是一個可以考慮採用的方法。本研究以Schoenfeld (1985) 的數學解題模式為理論基礎，並採用放聲思考法，探討國小學生的數學解題歷程。本研究樣本為六位國小六年級學生(男生3位，女生3位)。研究工具為「非例行性題目」。 研究結果發現： 1.在影響解題歷程的四個變項(資源、捷思、控制與信仰系統)方面，低數學能力組的兩位學生，與中、高數學能力組的四位學生，有明顯不同的解題行為。 2.在讀題與計畫這兩階段，不同數學能力組的六位學生，在階段間的轉換次數與使用的時間上，並沒有特別的不同。 3.在分析、探索、執行與驗證這四個階段，低數學能力組的兩位學生在階段間的轉換次數與使用的時間上，皆比中、高數學能力組的四位學生少。

The research of mathematical problem solving had emphasized the result of problem solving, this method of research can't deeply discuss to the behavior of problem solving during the problem solving process. To improve this shortcoming, thinking aloud is a useful method. Based upon the Schoenfeld's mathematical problem solving model, the study adopts the thinking aloud method to explore the process of mathematical problem solving. The subjects of this research are six at the 6th grade students(three boys and three girls) at elementary school. The research instrument is "non-routine mathematical problem test". The major results of this research as follows: 1.As for the four variables affecting the process of problem solving( resources, heuristics, control, and belief system), the two students of low mathematical achievement group havesignificant difference of problem solving behavior with the four students of medial and high mathematical achievement group. 2.Both on the reading and planning phrases, the students with different mathematical achievement have no significant differences on the transitional number and the time using. 3.On the analysis, exploration, implementation, and verification phrases, the two students of low mathematical achievement group have less significance on the transitional number and time using than the four students of medial and high mathematical achievement group.