This study has two parts: phase I designed activities to support all students’ learning preferences, and phase II used open-ended questions and activities based on these preferences to develop students’ mathematical critical thinking skills in polynomials at all performance levels (i.e., high-achieving, fair-achieving, and low-achieving students). This research used an embedded mixed-method design. The subjects selected were 28 out of 98 seventh graders at a boys’ junior high school in Bangkok, Thailand, who were chosen by cluster random sampling technique. The instruments, which were validated by five experts, included a questionnaire, lesson plans, exit tickets, interview protocols, and tests of critical thinking skills in polynomials. The content validity was assessed via expert judgment, and reliability was assessed by item analysis. The quality and effectiveness of the instruments were acceptable. The research results showed the following: (1) most students at all performance levels prefer activities in which they can learn from participating in classroom activities, such as games, activities with real-life applications, and activities involving listening instead of reading and writing, and (2) critical thinking skills in high-achieving and fair-achieving students were at the fair level, while those of low-achieving students were poor. Analysis was the highest critical thinking subskill among high-achieving and low-achieving students, while interpretation was the highest subskill in among fair-achieving students. Open-ended questions and activities based on students’ preferences appear to be practical for developing critical thinking skills among students of all achievement levels.