The aim of this study is to examine the mathematical problem-solving beliefs and problem-solving success levels of primary school teacher candidates through the variables of academic success and gender. The research was designed according to the mixed methods technique in which qualitative and quantitative methods are used together. The working group, comprised of 138 freshman students studying in the Primary School Teaching Department at a state university formed the quantitative data of the research. Using criterion sampling, a purposeful sampling method technique, 36 students were identified for forming the qualitative data group. The Belief Scale Regarding Mathematical Problem Solving, as developed by Kloosterman and Stage; the Identifying Test of Problem-Solving Success Levels, written by the author; and a semi-structured interview form were used to collect the data. The data was analyzed by MANOVA and two-way ANOVA testing respectively for the quantitative dimensions of the study while qualitative data was analyzed through the descriptive analysis method. Research results concluded that there was not a significant difference between the belief levels of teacher candidates with high and low problem-solving success levels for the dimensions of mathematical skill, the place of mathematics, and problem-solving skills in the Belief Scale Regarding Mathematical Problem Solving. Significant differences were found, however, in the sub-dimensions of understanding the problem and the importance of mathematics, regarding teacher candidates with high problem-solving success levels. It was also clearly seen from the results of the research that the ideas of teacher candidates with low and high problem-solving success levels were similar to each other. Based on these results, it is suggested that importance be given to classroom activities that positively affect the beliefs of primary school teacher candidates in regard to problem solving and learning mathematics. Also, researches using experimental designs can be performed by controlling the interaction of variables that can affect problem-solving beliefs.