Due to the increased use of scientific models and modelling in K-12 education, there is a need to uncover its effects on students over time. Prior research has shown that the use of scientific modelling in K-12 classes is associated with improved conceptual knowledge and problem-solving skills. However, few studies have explicitly tested the longitudinal benefits of using model-based instruction on students’ scientific reasoning skills (SRS) and content knowledge. This paper studies the effects of the use of modelling-based pedagogy in a longitudinal comparative case study on students’ SRS using hierarchical linear modeling. Our findings showed that initial exposure to modelling-based instruction increased scientific reasoning scores significantly. By the end of their first year of science instruction, the average high school freshman in our study achieved the scientific reasoning level of many undergraduate STEM majors. More importantly, students in the lowest quartile of scientific reasoning demonstrated increased scores over the three years of the modeling-based course sequence. In addition, reasoning scores in the modelling classes were a significant predictor of post-content knowledge in all subjects. Our results suggested that students should be exposed to model-based instruction early and consistently to achieve equity in science instruction.