Creative problem-finding ability (CPFA) is an important component of creativity, but research into it has just started and results so far could not offer much guidance to teaching. This study utilized a 2 (teaching materials or tasks of different difficulty: high and low) × 3 (group member construction: homogeneous group, heterogeneous group, and voluntary group) between subjects design to examine how the group member construction affected student's peer interaction, and then affected creative problem-finding ability during a long-term classroom teaching process. Before the experiment, 217 5th-grade students (108 boys, 109 girls) in 4 classes were randomly chosen from 1 primary school in China to take the CPFA Test and then, according to the score on this, 188 students were chosen and matched into 6 equal groups. All of the participants in 6 groups received 16 classes of a thinking intervention program in 16 weeks and did the CPFA Test and Peer Interaction Ability Test after the experiment. The results showed that: (a) Teaching materials (or tasks) of different difficulty had no significant impact on improving students' creative problem-finding ability, (b) group member construction had a significant influence on students' peer interaction and CPFA, (c) Students of different initial CPFA in different groups had a significant difference on their peer interactive level and improvement of their CPFA, and (d) peer interaction had a positive correlation with students' CPFA.