This study examined the effects of a Problem-Based Learning (PBL) model oriented to Balinese local wisdom and students’ cognitive styles on misconceptions and scientific literacy. A quasi-experimental design with a 3×2 factorial arrangement was employed, involving three learning models (PBL with local wisdom, standard PBL, and conventional learning) and two cognitive styles (reflective vs. impulsive). The sample consisted of 178 fifth-grade students from 12 schools, selected through a multi-stage random sampling process. Data were collected using a misconception test (α = 0.92), a scientific literacy test (α = 0.79), and the Matching Familiar Figures Test (α = 0.72). Results of two-way MANOVA showed significant main effects of learning model (Wilks’ Λ = 0.697, F (4,342) = 16.93, p < 0.001, η² = 0.16) and cognitive style (Wilks’ Λ = 0.407, F (2,171) = 124.7, p < 0.001, η² = 0.59). A significant interaction was also observed between learning model and cognitive style (Wilks’ Λ = 0.859, F (4,342) = 6.73, p < 0.001, η² = 0.07). Reflective students benefited most from PBL with Balinese local wisdom, showing the lowest misconceptions and highest scientific literacy, while impulsive students performed relatively better under conventional learning. These findings emphasize the importance of integrating cultural context and cognitive diversity into science education.