The purpose of this study was to investigate whether learners with different science content knowledge backgrounds, namely physics and science education graduates, construct models differently in the same computer programming environment with graphically represented program language and for the same subject matter (1D collisions). To do so, we selected 28 participants for each group and offered them the same modeling-based learning treatment. Data collection involved the administration of two paper-and-pencil tests, the participants’ created models, and screen-capture data (both video and sound). The first test examined the participants’ content knowledge on 1D collisions and the second one participants’ modeling competence. The data analysis involved both qualitative and quantitative methods. The findings revealed that variation in science background knowledge appears to affect the learners’ modeling competence, the types and nature of the models created, and the model creation progression followed.