We evaluated visual and vestibular contributions to vertical self motion perception by exposing subjects to various combinations of 0.2 Hz vertical linear oscillation and visual scene motion. The visual stimuli presented via a head-mounted display consisted of video recordings of the test chamber from the perspective of the subject seated in the oscillator. In the dark, subjects accurately reported the amplitude of vertical linear oscillation with only a slight tendency to underestimate it. In the absence of inertial motion, even low amplitude oscillatory visual motion induced the perception of vertical self-oscillation. When visual and vestibular stimulation were combined, self-motion perception persisted in the presence of large visual-vestibular discordances. A dynamic visual input with magnitude discrepancies tended to dominate the resulting apparent self-motion, but vestibular effects were also evident. With visual and vestibular stimulation either spatially or temporally out-of-phase with one another, the input that dominated depended on their amplitudes. High amplitude visual scene motion was almost completely dominant for the levels tested. These findings are inconsistent with self-motion perception being determined by simple weighted summation of visual and vestibular inputs and constitute evidence against sensory conflict models. They indicate that when the presented visual scene is an accurate representation of the physical test environment, it dominates over vestibular inputs in determining apparent spatial position relative to external space.