Resumen: A Content and Language Integrated Learning (CLIL) method, in which some subjects are taught in a foreign language (English), was initiated in Spain in 2005 and has progressively extended to half of public schools. The results have been very positive; however, it has been argued that studying subjects in a foreign language may reduce the educational outcomes of students. This paper evaluates this criticism in the Madrid bilingual program adopting a Propensity Score Matching approach, with the PISA 2015 and 2018 data. The model defines a homogenous student subsample from bilingual and nonbilingual schools in terms of the observable characteristics that may jointly influence both the selection of school type and educational scores. Our results, robust to the sample and unobservable hidden bias, indicate that the Madrid bilingual program, in addition to improving students' English level, does not reduce the skills of subjects taught in English or in Spanish. Idioma: Inglés DOI: 10.1016/j.stueduc.2024.101386 Año: 2024 Publicado en: Studies in Educational Evaluation 83 (2024), 101386 [11 pp.] ISSN: 0191-491X Financiación: info:eu-repo/grantAgreement/ES/DGA/S23-20R Financiación: info:eu-repo/grantAgreement/ES/DGA/S23-23R Tipo y forma: Artículo (Versión definitiva) Área (Departamento): Área Economía Aplicada (Dpto. Economía Aplicada)