Several studies have shown that learning mathematics can be more effective if students solve problems by their own reasoning than if they follow given solution methods. However, when a student needs help with solving a task it is more challenging for the teacher to support the student’s own reasoning than to describe the method. In a collaborative research and development project, teachers and researchers developed a framework for diagnosing in- the-moment problem-solving difficulties. The present study compares the characteristics of experiment and control group teachers’ diagnoses of student difficulties. The study included 83 teachers making 298 diagnoses. The results show that the teachers in the experiment group, to a higher degree, focused on the problem-solving process, while teachers in the control group focused more on aspects of student competence. The teachers in the experiment group also anchored their diagnoses in the students’ statements and written notes to a greater extent than the control group teachers.