Activation of the vestibular system can either increase or decrease ventilation. The objectives of the present study were to clarify whether these different responses are the result of activating different vestibular subnuclei, by addressing three questions. Do neurones within the medial, lateral and spinal vestibular nuclei (VN_M, VN_L and VN_S, respectively) function differently in respiratory modulation? Is the ventral medullary nucleus gigantocellularis (NGC) required to fully express the VN-mediated respiratory responses? Is glutamate, by acting on N-methyl-D-aspartic acid (NMDA) receptors in the vestibular subnuclei, capable of modulating respiration? In anaesthetized, tracheotomized and spontaneously breathing rats, electrical stimuli (< 10 s) applied in the VN_L and VN_S significantly elevated ventilation by 35 % and 30 % (P < 0.05), respectively. However, VN_M stimulation produced stastically significant (P < 0.05) changes that differed depending upon the stimulation site: either ventilatory inhibition (by 40 % in 57 % of the trials) or excitation (by 55 % in 43 % of trials), and which were often accompanied by a pressor response. These electrical-stimulation-evoked cardiorespiratory responses were almost eliminated following microinjection of ibotenic acid into the stimulation sites (P < 0.05) or bilaterally into the NGC (P < 0.05). As compared to vehicle, microinjection of NMDA into the unilateral VN_M, VN_L and VN_S significantly increased ventilation to 74 %, 58 % and 60 % (P < 0.05), respectively, with no effect on arterial blood pressure. These data suggest that neurones within the vestibular subnuclei play different roles in cardiorespiratory modulation, and that the integrity of the NGC is essential for the full expression of these VN-mediated responses. The evoked respiratory excitatory responses are probably mediated by glutamate acting on NMDA receptors, whereas the neurotransmitters involved in VN_M-mediated respiratory inhibition and hypertension remain unknown.