The purpose of this research is to evaluate the type and accuracy of the visual representations that the pre-service teachers generate in the pedagogical solution of the word problems. The data on participants' spatial competence levels were obtained through Purdue Spatial Visualization Test and the data on the type of visual representations were obtained by coding the data obtained from the solution of a word problem given to the participants. In the analysis of the data, descriptive analysis and content analysis were applied. As a result of the research, the visual representation types (schematic and pictorial) generated by the pre-service teachers have been determined. The results suggest that the pre-service teachers with high spatial skills can generate the structured schematic representations correctly whereas those with low spatial skills have tendency to generate pictorial representations. Some suggestions for teacher education are presented.