How much mathematics does one need to have to be a good problem solver? This has been an ongoing debate. One school of thought believes the more substantive the mathematics knowledge, the better one is at problem solving. This paper proposes that while mathematics knowledge is important, another construct is equally if not more important - belief of self as a problem solver. Lack of confidence can be a major obstacle in problem solving. This paper reports findings of a study which sought to test the latter belief. Forty-eight trainee teachers in the final semester of a four-year degree primary teacher education programme participated in this study. In addition to three mathematics method modules, these trainee teachers had also completed two modules specifically introduced to improve their mathematics content knowledge. The trainee teachers were asked to solve two questions which required the application of fraction concepts specific to primary mathematics curriculum. Besides solving the two problems, the trainee teachers were asked to report their perceptions of themselves as problem solvers. Seven volunteers were interviewed to provide more in-depth knowledge of their problem solving behaviours. Despite the fact that they had been taught more mathematics, the self-reporting data found that the trainee teachers were highly influenced by their belief of themselves as problem solvers. The interview data suggested that trainee teachers appreciated activities where they were given a chance to discuss how to solve problems and also how to teach such problem solving skills.