Common ground (CG), i.e. the knowledge, beliefs and assumptions that interlocutors mutually share in interaction, is fundamental t o successful communication (Clark, 1996). An increasing number of studies have shown that speakers use co - speech gestures at the same rate (or even higher) when they share CG as opposed to when they do not (e.g. Campisi & Ozyurek, 2013; Holler & Wilkin, 20 09; Holler, Tutton & Wilkin, 2011). There are two alternative explanations for this finding. On the one hand, it has been argued that mentally representing our addressee’s knowledge can require considerable cognitive effort (Pickering & Garrod, 2004). Henc e, gesture rate may be high in CG contexts because the cognitive effort involved in mentally representing CG is considerable. In contrast, this high gesture rate may be due to the fact that gestures play an important communicative role, even when conveying information that is already mutually shared (Holler & Wilkin, 2009). The present study tested these two hypotheses by combining the manipulation of CG with a manipulation of communicative context. We used a 2(CG) x 3(communication context) between - partici pants design (18 participants per condition, N=108). All participants watched a short film and narrated it to their addressee. Addressees had either seen parts of the film together with the speaker (CG) or not (no - CG). In addition, we manipulated communica tive context by asking speakers to narrate their story either face - to - face, via an occluding screen, or into a tape - recorder, a manipulation that has been shown to affect gesture rate in no - CG contexts (Bavelas et al., 2008). Our results revealed a signifi cant main effect of communicative context, with gesture rate being highest in the face - to - face condition, followed by the screen condition, and lowest in the tape - recorder condition. Importantly, we did not find a main effect of common ground on gesture ra te, and no interaction between our two factors. This finding supports the hypothesis that gestures representing CG information are communicatively intended as opposed to being triggered by an increased cognitive load.