Kemampuan pemahaman konsep sangat dibutuhkan dalam proses pembelajaran matematika yang harus dikuasai untuk menunjang keberhasilan proses pembelajaran. Penelitian kuantitatif ini mengkaji kemampuan pemahaman konsep matematis siswa dengan mengimplementasikan model VARK-Fleming dan self-concept sebagai variabel kovariat. Penelitian ini bertujuan untuk mengetahui pengaruh model VARK-Fleming terhadap kemampuan pemahaman konsep matematis dengan mengontrol self-concept siswa dan pengaruh variabel kovariat self-concept terhadap kemampuan pemahaman konsep matematis siswa, serta pengaruh secara simultan model VARK-Fleming dan self-concept terhadap kemampuan pemahaman konsep matematis siswa. Partisipan dalam penelitian ini berjumlah 64 siswa SMP, dengan rincian 32 siswa belajar menggunakan model VARK-Fleming dan 32 siswa belajar dengan model ekspositori. Alat yang digunakan untuk mengumpulkan data adalah angket dan tes uraian. Pengujian hipotesis menggunakan analysis of covariance (one-way ancova). Hasil penelitian ini menunjukkan bahwa terdapat pengaruh lebih baik model pembelajaran VARK-Fleming dibandingkan model ekspositori dan self-concept terhadap kemampuan pemahaman konsep matematis siswa baik secara parsial maupun simultan. Concept understanding is required for the learning process to be successful. This quantitative study investigates students' mathematical concept understanding abilities using the VARK-Fleming model and self-concept as a covariate variable. The purpose of this research is to determine the effect of the VARK-Fleming model on students' ability to understand mathematical concepts by controlling for students' self-concept and the effect of the covariate variable of self-concept on students' ability to understand mathematical concepts, as well as the simultaneous concept of the VARK-Fleming model and self-concept on students' ability to understand mathematical concepts. Sixty-four junior high school students participated in this research, with 32 learning the VARK-Fleming model and 32 learning the expository model. Questionnaires and descriptive tests were employed to collect information. The analysis of covariance was utilized for hypothesis testing (one-way ANCOVA). The findings of this research reveal that the VARK-Fleming learning model has a better effect on students' mathematical concept understanding capacity than the expository model and self-concept, both partially and concurrently.