Abstract If multimodal work in terms of Construction Grammar is still rare nowadays, this is not just because the upsurge of multimodality in linguistics in general is rather recent. Attempts to include non-verbal layers of expression (especially gesture) in Construction Grammar have uncovered a number of theoretical issues that need to be reflected upon before any serious claims on the existence of multimodal constructions can be made. While some scholars take these issues as reasons for leaving the non-verbal outside of the scope of Construction Grammar, this paper shows that several of these issues are not actually related to multimodality, but rather hint at more general theoretical issues in Construction Grammar. Hence, it is argued that these issues should be seen as an incentive for rethinking and refining the notion of ‘construction’, rather than as a reason for leaving multimodality aside.