Serratrice et al. (2004) propose to extend Hulk & Müller’s (2000, 2001) hypothesis on cross-linguistic influence in early child bilingualism to include cases of influence after instantiation of the C-system (i.e. at a later stage of development). In the present article, we explore whether such an extension can successfully account for the use of dislocation, a topicmarking device, in French-English and French-Dutch bilingual children. On the one hand, our results support the extended formulation of the model: we find cross-linguistic influence in the bilingual data as predicted. On the other hand, certain aspects of our results cannot be sufficiently accounted for under the extended formulation. We discuss several other factors which may interact with those cited in Hulk & Müller’s model, such as input frequency, transparency of syntactic-pragmatic mapping, complexity of syntactic structures, and Chomskyian economy, which may need to be considered in future research on cross-linguistic influence.