This study aims to describe the students' mathematical problem solving ability on the material of building space based on the characteristics of students' thinking. This type of research is descriptive qualitative research. Sample selection using purposive sampling technique. Data collection was carried out by distributing questionnaires on the characteristics of students' thinking, problem-solving ability tests and interviews. The data analysis technique uses the Miles and Huberman model and uses technical triangulation in the validation test. The results showed that students with the characteristics of concrete sequential thinking had met the four problem solving indicators, but students were more dominant in the stage of understanding the problem. Students with the characteristics of abstract sequential thinking also fulfill four problem-solving indicators and are dominant in the stage of developing a problem-solving plan. Students with the characteristics of abstract random thinking fulfill four problem solving indicators but need to take a long time to solve them.