The present study falls within the framework of an interdisciplinary project on Content and Language Integrated Learning (CLIL) in French-speaking Belgiumi. It explores the effects of input on written proficiency in immersive (CLIL) and non-immersive (non-CLIL) language learning settings using various input measures. Several studies (e.g. Zydatiß 2007; Lorenzo & Moore 2010; Jexenflicker & Dalton-Puffer 2010; Gené-Gil, Juan Garau & Salazar-Noguera 2015; Martínez 2015; Bulon, Hendrikx, Meunier & Van Goethem forthcoming) compare the language proficiency of learners in CLIL and non-CLIL settings using global measures of complexity, accuracy and/or fluency, typically referred to as CAF (Housen, Kuiken & Vedder 2012; Norris & Ortega 2009). However, few studies on CLIL have controlled for the possible effect of L2 exposure (Saladrigues & Llanes 2014), while time spent on learning a second/foreign language has been recognized as one of the most important factors for successful acquisition. The amount and quality of target language (TL) input has become a focus of interest for many SLA researchers (e.g. Kinsella 2009 and Moyer 2009 on formal and informal contact; Llanes & Muñoz 2009 and Pérez-Vidal & Juan-Garau 2011 on study abroad; Long 1983 on formal instruction; Johnstone 2007 on length of instruction). The present study seeks to investigate 1) the written proficiency of CLIL and non-CLIL secondary school pupils and 2) the contribution/impact of various input measures to explain potential differences in proficiency between immersive and non-immersive pupils. Since CLIL programs aim to provide more target-like and input-rich environments than non-CLIL programs, we can expect a more native-like acquisition of the target language, i.e. a more native-like level of syntactic complexity and lexical diversity in the writing of CLIL pupils. The analysis is based on a corpus of written data in the form of 378 e-mails on similar topics. The participants are 5th year French-speaking ...