We analyze the effect of substituting a weekly mathematics lessons in primary school grades 1-3 with a lesson in mathematics based on chess instruction. We use data from the City of Aarhus in Denmark, combining test score data with a comprehensive data base from administrative register. We use a difference-in-differences approach to investigate treatment effects on the treated and tend to find positive effects. Looking at sub groups, we find significant positive effects for native Danish children, while we find no effects for children of immigrants.