Research points to the positive impact that gender-segregated schooling and classroom initiatives exert on academic attainment. An evaluation of these studies which reveal positive effects highlights, however, that students are typically selectively assigned to single- or mixed-gender instructional settings, presenting a methodological confound. The current study controls for students’ prior attainment to appraise the efficacy of a single-gender classroom initiative implemented in a co-educational high school in the United Kingdom. Secondary data analysis (using archived data) was performed on 266 middle-ability, 11–12 year-old students’ standardized test scores in Languages (English, foreign language), STEM-related (Mathematics, Science, Information and Communication Technology), and Non-STEM subjects (art, music, drama). Ninety-eight students (54, 55% female) were taught in single-gender and 168 (69, 41% female) in mixed-gender classrooms. Students undertook identical tests irrespective of classroom type, which were graded in accordance with U.K national curriculum guidelines. Controlling for students’ prior attainment, findings indicate that students do not appear to benefit from being taught in single-gender relative to mixed-gender classrooms in Language and STEM-related subjects. Young women benefitted from being taught in mixed-gender relative to single-gender classes for Non-STEM subjects. However, when prior ability is not controlled for, the intervention appears to be effective for all school subjects, highlighting the confounding influence of selective admissions. These findings suggest that gender-segregated classroom initiatives may not bolster students’ grades. It is argued that studies that do not control for selection effects may tell us little about the effectiveness of such interventions on scholastic achievement.