The purpose of this study was to investigate the effectiveness of cognitive conflict based physics instruction over traditionally designed physics instruction on preservice primary school teachers at grade 2. The subjects were 82 (27 boys, 55 girls) second grade pre-service teachers in two classes. One of the classes (42 students) was randomly assigned as experimental and the other class (40 students) assigned as control group. Both groups were taught by the same instructor. While the experimental group received cognitive conflict based physics instruction, control group were taught by traditionally designed physics instruction. The data were obtained through Thermal Concept Evaluation test (TCE). Prior to instruction, students in both groups were pre-tested by TCE in order to determine their initial understanding of heat and temperature at the beginning of instruction. The same tests were applied as posttest after the instruction. Independent samples t-test on pre-test scores showed that there was no statistical significant difference between experimental and control group at the beginning of the instruction in terms of understanding of heat and temperature concepts. ANCOVA results showed that mean scores on the post-TCE of students in experimental group were significantly higher than those of the control group. While interaction between gender difference and treatment made a significant contribution to the variation in achievement, gender difference did not.