Metacognition is the ability to think about thinking. One of the learning models in geometry is the van Hiele model, which consists of learning phases and geometry thinking level. However, geometry learning strategy in Indonesia does not stress metacognition and geometry thinking level. Hence, this study aims to examine the effectiveness of geometry learning strategy based on the infusion of metacognition in van Hiele model compared to van Hiele learning phases in helping secondary school students to improve their geometry thinking level. The quasi-experimental study was conducted six-week with 90 students. The students selected purposively divided into two groups, with 30 students in both treatment groups, respectively. The instrument employed van Hiele Geometry Test (vHGT) before and after the treatment to measure the student's geometry thinking level. Data were in ordinal form analyzed descriptively and inferentially using Mann-Whitney U. The result revealed the significant difference between the final geometry thinking level in both groups. Thus, it can be concluded that the geometry learning strategy based on the infusion of metacognition in van Hiele model is more effective in improving the student's geometry thinking level than the geometry learning strategy based on van Hiele model.