This article reports a mixed-methods study that examined academic success in an Economics programme at a public university in Turkey. Test score data from English Medium Instruction (EMI) and Turkish Medium Instruction (TMI) courses, and general English proficiency (GEP) scores were collected from fourth-year students (n = 159). Follow-up semi-structured interviews were conducted with 12 students. Results showed that General English Proficiency was not a statistically significant predictor of EMI academic success. TMI academic success, however, did significantly predict success in EMI. This illustrates that EMI success is better augmented by students taking some courses through their native language alongside EMI courses. A Multilingual Model (a hybrid rather than a purist model) of EMI implementation is therefore supported (Macaro, 2018:233). Qualitative data supported this as students affirmed the mediatory effect of first language courses on their EMI academic success. Suggestions for future research as well as practical pedagogical implications are provided. (C) 2020 Elsevier Ltd. All rights reserved.