This chapter presents the current state of research in multimodal Construction Grammar with a focus on co-speech gestures. We trace the origins of the idea that constructions may have to be (re-)conceptualized as multimodal form–meaning pairs, deriving from the inherently multimodal nature of language use and the usage-based model, which attributes to language use a primordial role in language acquisition. The issue of whether constructions are actually multimodal is contested. We present two current positions in the field. The first one argues that a construction should only count as multimodal if gestures are mandatory parts of that construction. Other, more meaning-centered, approaches rely less on obligatoriness and frequency of gestural (co-)occurrences and either depart from a recurrent gesture to explore the verbal constructions it combines with or focus on a given meaning, for example, negation, and explore its multimodal conceptualization in discourse. The chapter concludes with a plea for more case studies and for the need to develop large-scale annotated corpora and apply statistical methods beyond measuring mere frequency of co-occurrence.