This study aims at determining to what extent visuals contribute to success in non-routine problem solving process and what types of errors are made when solving with visuals. Comparative model was utilized for identifying the effect of visuals on student achievement, and clinical interview technique was used to determine the types of errors. In the study, 370 primary fourth-grade students were applied with the four non-routine problems; first in verbal form, then in visual form by changing only the numbers without changing the operational stages that should be used when solving the question. As a result, it was seen that the use of visuals decreased the number of unanswered questions by 11%, and increased the number of correctly answered questions by 12%. It was found at the end of the study that when students utilized the visuals they drew rather than the given visuals, this contributed to the problem solving achievement more. Another finding achieved in the study is that the students made errors originating from lack of knowledge, misinterpretation, incorrect structuring, incomplete structuring, and misplacement of what is given respectively.