A new computational linear algebra course was developed and offered at a large public university in the Midwest. The new course included applied computational materials but with the trade-off of less coverage of linear algebra topics than the pre-existing traditional linear algebra course. We studied the impact of this new computational linear algebra course on students’ exam performance in a subsequent numerical methods course, while controlling for students’ performance in prerequisite computer science and mathematics courses. Our results show that for students with lower mathematics proficiency (namely those who needed to take Calculus II at the university), the new computational linear algebra course improved their average exam performance in the subsequent course. For students with higher level of mathematics proficiency (namely those who passed out of Calculus II), no significant difference was observed between students from the computational and traditional courses.