This project was conducted as quasi-experimental design. We randomly assigned two classes of the third grade of the primary school to experimental group and control group, which take digital school network learning and classroom learning, respectively. The purpose of this project was to compare the learning effect and the distribution of the van Hiele level of geometric thinking through two different learning methods. Moreover, we wanted to further realize the solid geometry concept of the sampled students by a structured interview, and surveyed the learning attitude and opinion of the experimental group students using digital school network learning.