The learning process in elementary school found a lack of use of learning media and a lack of student interest in science lessons. This study aims to analyze the effectiveness of realia and video media on science cognitive learning outcomes. Experiments are used as a type of research with a quantitative approach. The design of this research is correlative. The population of this study was all fifth-grade students. Fifteen students were selected using purposive sampling the sample. Pre-test and post-test were given to determine cognitive learning outcomes. The data obtained were processed using paired sample t-tests because the data were average. In this study, it can be concluded that the use of realia media is more effectively applied in science learning to students' cognitive learning outcomes than video media. The acquisition of sig indicates this chapter. 0.020 < 0.05. Therefore, the initial hypothesis is accepted, and the alternative hypothesis is rejected.