Key points When standing and holding an earth‐fixed object, galvanic vestibular stimulation (GVS) can evoke upper limb responses to maintain balance. In the present study, we determined how these responses are affected by grip context (no contact, light grip and firm grip), as well as how they are co‐ordinated with the lower limbs to maintain balance. When GVS was applied during firm grip, hand and ground reaction forces were generated. The directions of these force vectors were co‐ordinated such that the overall body sway response was always aligned with the inter‐aural axis (i.e. craniocentric). When GVS was applied during light grip (< 1 N), hand forces were secondary to body movement, suggesting that the arm performed a mostly passive role. These results demonstrate that a minimum level of grip is required before the upper limb becomes active in balance control and also that the upper and lower limbs co‐ordinate for an appropriate whole‐body sway response. Abstract Vestibular stimulation can evoke responses in the arm when it is used for balance. In the present study, we determined how these responses are affected by grip context, as well as how they are co‐ordinated with the rest of the body. Galvanic vestibular stimulation (GVS) was used to evoke balance responses under three conditions of manual contact with an earth‐fixed object: no contact, light grip (< 1 N) (LG) and firm grip (FG). As grip progressed along this continuum, we observed an increase in GVS‐evoked hand force, with a simultaneous reduction in ground reaction force (GRF) through the feet. During LG, hand force was secondary to the GVS‐evoked body sway response, indicating that the arm performed a mostly passive role. By contrast, during FG, the arm became actively involved in driving body sway, as revealed by an early force impulse in the opposite direction to that seen in LG. We then examined how the direction of this active hand vector was co‐ordinated with the lower limbs. Consistent with previous findings on sway anisotropy, FG skewed the direction of the GVS‐evoked GRF vector towards the axis of baseline postural instability. However, this was effectively cancelled by the hand force vector, such that the whole‐body sway response remained aligned with the inter‐aural axis, maintaining the craniocentric principle. These results show that a minimum level of grip is necessary before the upper limb plays an active role in vestibular‐evoked balance responses. Furthermore, they demonstrate that upper and lower‐limb forces are co‐ordinated to produce an appropriate whole‐body sway response.