Online learning causes teachers to be considered more active than students, so students’ higher-order thinking skills decrease. Limited face-to-face learning is a new normal policy in the field of education. Learning models can be one way to recreate active learning to improve student’s learning outcomes. This study determines the effectiveness of the problem-based learning (PBL) model in enhancing student learning outcomes of geography in new normal learning. This research is a nonequivalent pretest-posttest control group design. The population is all class XI social study (IPS) at Senior High School 13 Banjarmasin, Indonesia. The sample is used as the experimental group, namely XI IPS 3 totaling 17 students, and the sample is used as the control group, namely XI IPS 2 totaling 14 students. Data collection techniques use tests, non-test, interviews, and documentation. The results are the independent t-test in the experimental and control groups showed that the PBL model effectively improved student learning outcomes in geography in the new normal learning era. The implication is the PBL model can be applied to new normal learning by teachers by elaborating with learning technology which is still adapted to the characteristics of students so that learning objectives are achieved.