本研究目的在探究不同多元智能分組以進行網路合作數學問題解決之效益，主要以多元智能進行學生分組，在研究者自行設計的課程網站中，採合作學習模式進行網路數學問題解決，以了解不同學習成就與不同多元智能分組狀況之學習成效。實驗組以學生的多元智能表現分組，控制組以學業成就進行分組，探討在此學習模式下，不同分組條件的學生在數學成就評量測驗以及合作學習歷程的差異情形，並進一步探討實驗組三類不同智能（個人、數理及其他智能）學生在合作學習歷程的表現情形。研究對象為國小六年級學童，實驗組有36人，控制組有34人，共計70人。採準實驗研究法，以t檢定等統計方法分析蒐集的資料，結果如下：1. 實驗組與控制組學生在「數學評量測驗」成績沒有顯著差異。2. 以多元智能分組的實驗組在分組作業中的「豐富性」及「創造性」顯著優於控制組。3. 實驗組學生在合作學習歷程評量的得分顯著優於控制組。 The main purpose of this study is to explore the benefit of dividing into groups with different multiple intelligence to proceed using cooperative learning through internet on problem solving of mathematics. Three groups were divided based on student personal intelligence, and then using cooperative learning model to proceed problem solving of mathematics through internet on a curriculum website. The students of experiment group were divided into groups by their multiple intelligence performance, and the students of control group were divided into groups with their learning achievement. The subjects of this study were seventy sixth-grades students; there were thirty-six in experiment group and thirty-four in control group. The Quasi-experimental design was used at the study. Several conclusions were presented as follows:1. The experiment group and the control group students made no remarkable difference on “Test of Math Achievement” grades.2. About the “abundance” and “creativity” of their grouping homework, the experiment group that divided into groups with multiple intelligence is obviously better than the control group.3. About the grades on the process of cooperative learning, the experiment group students are obviously better than the control group students.