We examined the vection induced by consistent and conflicting multisensory information about self-motion. Observers viewed displays simulating constant-velocity self-motion in depth while physically oscillating their heads left ^ right or back ^ forth in time with a metronome. Their tracked head movements were either ignored or incorporated directly into the self-motion display (as an added simulated self-acceleration). When this head oscillation was updated into displays, sensory conflict was generated by simulating oscillation along: (i) an orthogonal axis to the head movement; or (ii) the same axis, but in a non-ecological direction. Simulated head oscillation always produced stronger vection than `no display oscillation'öeven when the axis/direction of this display motion was inconsistent with the physical head motion. When head-and-display oscillation occurred along the same axis: (i) consistent (in-phase) horizontal display oscillation produced stronger vection than conflicting (out-of-phase) horizontal display oscillation; however, (ii) consistent and conflicting depth oscillation conditions did not induce significantly different vection. Overall, orthogonal-axis oscillation was found to produce very similar vection to same-axis oscillation. Thus, we conclude that while vection appears to be very robust to sensory conflict, there are situations where sensory consistency improves vection.