A system of bony canals and membranous ducts in the inner ear, collectively called the vestibular system, contains sensory cells that rapidly transduce head motion. Information from these cells drives some of the fastest reflexes in the body such as the vestibulo-ocular reflex and vestibulo-collic reflex which are vital for maintaining gaze, balance and orientation. Of these cells, the Type I sensory hair cells are surrounded by a cup-like terminal (calyx) of the afferent neuron and possess a characteristic low voltage activated potassium conductance (gKL) on their basolateral surface. The apposition of the presynaptic hair cell membrane and the postsynaptic calyx creates a unique synapse morphology that limits the diffusion of ions entering the intermediate cleft space. Transmission between the hair cell and afferent neuron across this cleft space occurs through the release of neurotransmitters (quantal) or without conventional neurotransmitters (nonquantal). I developed a computational model of the Vestibular Hair Cell-Calyx (VHCC) to investigate the mechanism of nonquantal transmission. The VHCC model uses expressions for K+ and Na+ electrodiffusion in the cleft, Hodgkin-Huxley-like ion currents based on whole-cell recordings, and the cable equation to calculate potentials in the hair cell, cleft, afferent calyx and afferent fiber. Using the VHCC model I show that nonquantal transmission occurs through changes in cleft electrical and K+ potential and that it is a case of ephaptic (due to proximity) transmission. I have identified features in existing experimental data that support the idea of ephaptic transmission between the hair cell and afferent calyx: 1) phase of synaptic transmission, 2) retrograde transmission, 3) fast post synaptic currents. Based upon this work, I advance the hypothesis that electrical and K+ potentials change throughout the intercellular spaces of the tightly packed vestibular epithelium, not just within the VHCC synaptic cleft. If true, this may explain the speed with which vestibular afferents encode even small head motions.