Argumentation skill is a part of the critical thinking ability needed by students to be successful in learning science. This study aims to understand the effect of the implementation of 4 Step Jigsaw (4SJ) and original jigsaw learning models in rehearsing the argumentation skill of junior high school students in the concept of plant structure and function. This argumentation skill is further investigated in the lens of varied students’ level of confidence. This study employs the quasi-experiment factorial 2 x 2 with the type of Pretest Posttest Nonequivalent Control Group design. The population was the students from grade 8 SMPN 17 Batanghari with total of 106 students, consisted of 44 males and 62 females. Some 53 students were recruited to be the sample grouped into 2 classes, i.e., the class 8A as the class 4SJ and the class 8C as the original jigsaw class. Data were collected using a valid self-confidence questionnaire and a valid pretest and a posttest in argumentation. Data were analyzed using two-way Anova test and statistics descriptive techniques. The results show that the 4SJ students have higher average of argumentation skill score (77.77) than the jigsaw students (71.17). It was also factual that students with better self-confidence also have higher average of argumentation skill score (75.13) than those with lower self-confidence (73.81). The results of two-way Anova test emphasize that there is an effect of the utilization of the models on the students’ argumentation skill. There is also an effect of the students’ self-confidence on the students’ argumentation skill. Even though it was very weak, it can be said that there is a potential for the learning models and the self-confidence to have an interaction in causing the growth of the students’ argumentation skill. It can be concluded that appropriate learning models and suitable self-confidence may affect students’ learning outcomes.