Previous studies have provided evidence that multimedia support is an important factor in test performance. However, few studies have compared the effects of real visual and model visual on test performance. The purpose of this study was to determine the effects of the model visual and real visual supported test items on elementary students' test performances in math. For this purpose, a math test designed in three types: text only, text + model visual, and text + real visual. The tests were applied randomly to 99 fifth-grade students. The study was designed with a post-test control group as a quantitative-dominant mixed method research. It was concluded that visual support in elementary school students' math tests increased their test scores and that the use of real visuals in tests increased the students' test performances. As an insider observer, the views of a math teacher supported these two main conclusions. Based on the results, suggestions were put forward for practice and future research. (C) 2017 Elsevier Ltd. All rights reserved.