The effects of single-sex education are hotly contested, both in academic and policy circles. Despite this heated debate, there exists little credible empirical evidence of the effect of a U.S. public school's decision to offer single-sex classrooms on the educational outcomes of students. This study seeks to fill this hole. Using administrative records for third through eighth graders in North Carolina public schools, the paper finds evidence that the offering of single-sex mathematics courses is associated with lower performance on end-of-grade math exams, and finds no evidence that the offering of single-sex reading scores increases performance on reading exams. Robustness checks are conducted. While the mathematics results are robust to the checks, the reading results fail an important check, and the baseline reading results should be interpreted with this in mind. Evidence of significant heterogeneity in the effect across schools is also presented.