This study aims to determine the effectiveness of the problem-based instructional learning model based on blended learning on learning outcomes in terms of students' mathematical anxiety. The research was conducted at SMK Negeri 3 Gorontalo. The type of research is Quasi Experiment with treatment design by level 2 x 2 using two-way analysis test and Tukey test. The results showed that the problem-based instructional learning model based on blended learning was effectively applied to students with low mathematical anxiety. This is indicated by the learning outcomes of students who have low mathematical anxiety in problem-based instructional learning based on blended learning of 85.39, better than the learning outcomes of students who have low mathematical anxiety indirect learning, which is only 63.44. This result is also strengthened through the -test with a significant level of 5% and showing .