Aim Galvanic vestibular stimulation (GVS) allows for robust vestibular responses and is therefore the gold standard in neuroimaging studies focusing on the central vestibular system. Nevertheless, there are several confounding side-effects which hinder the delineation of pure vestibular responses. Besides stimulating vestibular hair cells and afferent fibers [1] , mastoidal GVS also evokes somatosensory, vagal and nociceptive responses. Furthermore, GVS induces reflectory head movements leading to motion artefacts. Aim of our fMRI study was to assess and control for the impact of these confounders by means of local anaesthesia, optimized head fixation, and a sham-GVS stimulation with different current intensities. Methods We examined 44 right-handed healthy subjects (21 F; mean age 28 years) in three sessions (interval length between session > 24 h) with different conditions: GVS with 3.5 mA with/without local anaesthesia, sham-GVS with 3.5 mA/1 mA, and GVS with/without head fixation (headfixation system: Pearltec, Switzerland). Structural and functional images were obtained in a clinical 3T scanner (Siemens Magnetom Verio) with a 32-channel head coil. The protocol included a GVS stimulation session consisting of a T2*-weighted EPI sequence (TR 2s, 31 ascending slices, 3 mm in-plane resolution, slice thickness 3.5 mm), a MPRAGE sequence, and a resting-state session. Data analysis was performed using SPM 12 (Version 6407 Wellcome Department of Imaging Neuroscience, UK) and the motion fingerprint toolbox (Wilke, Neuroimage 2012) after standard preprocessing using DARTEL-based normalization. Results were considered significant at FDR p 0.05 (corrected at cluster level). Results Pain perception during GVS decreased significantly after local anesthesia (NRS-11 reduction from 4.5 to 1.7, Wilcoxon signed-rank test, Z = −3.79, p = 0.00). GVS with lidocaine evoked responses in the vestibular network, predominantly in right OP2 and the medial cingulate cortex bilaterally, which were absent during sham stimulation. Without anesthesia, confluent activation cluster were found in anterior and posterior insula-opercular regions. Activation of the insula was also shown bilaterally during sham-GVS, with a predominant effect in area OP1 and area PF with a current intensity of 3.5 mA. Sham-GVS with 1 mA led to bilateral responses in area Fp1, the precentral gyrus and the left posterior cingulum. Motion finger print analysis in GVS with and without head fixation revealed head motion predominantly in the roll plane, leading to inter-slice motion artefacts in temporo-parietal-opercular regions. These artefacts were eliminated using a new head fixation system. Conclusion In this fMRI study, we evaluated and minimized the effects of several confounders in GVS. Head motion artefacts as well as nociceptive and somatosensory responses may hinder the localization of actual vestibular response patterns, particularly in the insula and parietal operculum. Our results highlight the importance of controlling for these side-effects and lead the way towards a precise definition of the human vestibular cortical network.