Abstract: This paper analyzes some problematic studies in the field of web-based autonomous learning. The problems includes that few papers introduce latest models in their theoretical framework; some researches focus too much on comparing students' scores, whereas improving scores is not the aim of China's implementing web-based autonomous learning. In a way autonomous learning itself is the purpose. Scores should not be seen as the major standard for the success of autonomous learning, qualitative methods may be more appropriate instead; and that some experimental researches have logical mistakes in experimental study, that is, different experimental groups did not receive equal treatment.