The purpose of this study was to investigate the learning effects of flipped classrooms on elementary school and junior high school students in Taiwan. The data were analyzed according to the method of Hedges and Olkin (1985) and other technology of meta-analysis. The keywords “flipped classroom,” “flipped teaching,” and “flipped learning” were used to search for relevant theses in the Taiwanese Master’s thesis system. A total of 49 relevant theses, including 48 in the cognition domain and 31 in the affective domain, were obtained. The theses were analyzed on the overall weighted mean effect size (Hedges´ g), homogeneity by using Cochrane’s Q, and publication bias by using a funnel plot and the trim and fill method. Categorical model analysis was performed on the theses by using the Comprehensive Meta-Analysis version 3.0 software, which was designed by Borenstein, Hedges, Higgins, and Rothstein. The conclusions of this study are as follows: (1) Flipped classrooms positively influence learning effects for elementary school and junior high school students in Taiwan. The effect size for the cognition domain was 0.308, which equaled Cohen’s small effect. The effect size for the affective domain was 0.517, which equaled Cohen’s medium effect. (2) Moderator variables influence the learning effects of flipped classrooms on elementary school and junior high school students in Taiwan. Moderator variables of the learning effects of the cognitive domain include subjects, types of e-learning platforms, self-learning materials, self-learning inspection mechanisms, student-centered instructions, and elapsed teaching time. Moderator variables of the learning effects of the affective domain include students of different grades, subjects, types of e-learning platforms, self-learning materials, self-learning inspection mechanisms, student-centered instructions, and elapsed teaching time. According to the obtained results, some suggestions are proposed for elementary school and junior high school teachers, flipped classroom researchers, and meta-analysis researchers.