Self-assessment is essential to scientific literacy as stated by the National Research Council Committee on Conceptual Framework for the New K-12 Science Education Standards and has since been incorporated into the Next Generation Science Standards. However, little empirical evidence documents which instructional tools are beneficial in improving students' self-assessment in science learning. As such, we conducted a classroom-based quasi-experiment to test whether contrasting case-based instructional supports improved 390 introductory high school physics students' accuracy in self-assessment and academic performance when solving physics problems. We compared the effects of providing students with content knowledge and contrasting cases (contrasting good and poor solution) on students' physics problem solving and their ability to self-assess, with the effects of either presenting them with content knowledge and good solutions or teaching only content knowledge. We found contrasting examples improved the accuracy of students' self-assessments, promoted content learning, and enhanced the development of self-assessment strategies.