This paper presents results from a randomized experimental design replicated over four semesters that compared students' performance in understanding landform evolution processes as measured by the pretest to posttest score growth between two treatment methods: an online interactive simulation tool and a paper-based exercise. While both methods were shown to be effective at enhancing students' learning of the landform concepts and processes, there was no statistically significant difference in score growth between the two instructional methods. However, the attitudinal survey indicated that students consistently favored the simulation approach over the paper-based exercise. With the simulation method, female students showed greater score growth than males, especially for test items requiring higher level thinking. This indicates that the visually rich interactive simulation tool may be integrated to better support female students' learning in geoscience. Science major students generally outperformed non-science major students in terms of score growth, which suggests that background knowledge played an important role in realizing the potential of computer modeling in enhancing students' learning. Sufficient scaffolding is necessary to maximize the effect of interactive earth surface modeling in geoscience education.