We measured observers' perceived direction of self-motion resulting from the simultaneous presentation of visual and vestibular information, each simulating a different direction of motion. Sakurai et al. (2003) reported that when observers are presented vestibular stimulation consistent with rightward/leftward motion and visual stimulation consistent with forward/backward motion, they perceive themselves as moving in a direction intermediate to the directions specified by visual and vestibular information. Due to technical limitations, Sakurai et al. only employed one level of vestibular stimulation consistent with rightward/leftward motion, and the rate of expansion/contraction of the visual stimuli was not precisely controlled. To extend their study, we explored multiple levels of vestibular stimulation consistent with different motion trajectories, and systematically varied the rate of expansion/contraction of the visual stimuli using computer-graphics. In experiment 1, we measured the apparent direction of self-motion during oscillations on the swing while the observer was seated at various orientations relative to the objective trajectory of the swing motion and viewed expanding/contracting optic flow consistent with forward/backward self-motion via a head-mounted display. The frequency and maximum displacement (amplitude x 2) of the swing were constant at 0.33Hz and 60 cm respectively. Observers performed a rod-pointing task to report the perceived direction of self-motion. Only two of five observers perceived intermediate diagonal self-motion when the discrepancy between the direction of motion specified by optic flow and that specified by vestibular input was less than 90 degrees, meaning visual information affected the perceived direction of self-motion. Others, however, reported the veridical direction of their body motion ignoring the visual input. One possible reason for this failure to fully replicate Sakurai et al. is the magnitude of the vestibular signal was too large relative to the visual signal. Therefore, in experiment 2, we measured the apparent direction of self-motion again, reducing the displacement of the swing to 10 cm and the direction to only rightward/leftward, but systematically varying the visual information. Optic flow consisted of concentric sine wave rings (0.084 cpd) whose spatial phase oscillated between 0 (no motion), +/-270, and +/-540 degrees in order to generate different magnitudes of visual expansion/contraction independently of vestibular information. Most observers reported perceived directions intermediate to those specified by visual and vestibular information, that is, forwards and to the right (left) when the visual expansion was synchronized with the rightward (leftward) body motion. However, the perceived directions of self-motion did not clearly vary with the magnitude of optic flow. These results suggest that integration of visual and vestibular information is a weighted combination of both inputs when the discrepancy between them is small and their strengths are appropriately balanced.