This research explored how different models of Web-based dynamic assessment in remedial teaching improved junior high school student learning achievement and their misconceptions about the topic of "Weather and Climate." This research adopted a quasi-experimental design. A total of 58 7th graders participated in this research. Participants were divided into the experiment group (28 participants) and the control group (30 participants). The experiment group took the "personalized Web-based dynamic assessment" while the control group took the "non-personalized Web-based dynamic assessment." The two groups of students received the same lessons via conventional teaching prior to the experiment, and after the lessons were completed, they both took the pre-tests of the achievement test and the two-tier diagnostic test. After the tests, different models of Web-based dynamic assessment were administered to facilitate student learning. Once the Web-based dynamic assessments were accomplished, the two groups of students took the post-tests of the achievement test and the two-tier diagnostic test. Research findings reveal that students in the experiment group experience significantly better improvement in their learning achievement and misconceptions.