The purpose of this study was to examine the effectiveness of instructional strategies in associated with students' metacognition level in order to improve understanding in asynchronous web based discussion by comparing and analyzing the content of discussion. In experiment, sixty college students were divided into smaller groups, After that, student-student question generation strategy was applied to one class and student-teacher question generation strategy to the other class. Primary data sources were acquired in the transcripts of all students writing uploaded to the electronic discussion board. To analyze the types of reflection in the students' messages, Hatton and Smith's(1995) categories of reflective writing(descriptive, dialogic, critical reflection) were used to code all students' messages. Based on the statistical analysis of this examination, the results of this study were as follow; 1. the student-teacher question-generation strategy was more effective in enhancing descriptive, dialogic, critical reflection types. 2. student-teacher question-generation group with low metacognition students showed the more dialogic, critical reflection types. The implication of this study is that question-generation strategy with student-teacher interaction played useful on reflection and understanding to students with the low level of metacognition. Also, teachers have to interact with their student actively and intervene in web based discussion process.