While the factor ‘language contact’ is often named among the most important for the development of individual language attrition, empirical validations of this claim are few and far between. This contribution argues that a bilingual’s use of the first language comprises very diverse situations which cannot be subsumed under one predictor variable. Grosjean’s (2001) framework of language modes is invoked as a useful way of structuring the use of the L1 by immigrants. A statistical investigation of these different types of L1 use on the one hand and language proficiency data on the other demonstrates that the impact of both active and passive exposure to the first language on attrition is anything but straightforward.