Like all evoked potentials, where time-locked stimulus averaging is involved then the number of trials is only indirectly relevant. The metric that is directly relevant is residual noise – this these needs to be low enough for to make a conclusive interpretation of the results and, usually, more trials produce a lower residual noise.
The cVEMP is a relatively large amplitude evoked potential so, even with a relatively low number of trials the signal-to-noise ratio is often high enough to make a conclusive interpretation. The default number of sweeps set in the Interacoustics Eclipse instrument software is 200 and under the vast majority of circumstances this should be sufficient to achieve a suitably low residual noise in the averaged trace (i.e. to make interaural comparisons, to assess latency and amplitude and to differentiate “response present” and “response absent”).
As mentioned, although these concepts are transferable to other forms of evoked potential the cVEMP is a special case in one sense, and that is that the response relies on muscle contraction by the individual actively flexing their sternocleido mastoid muscle. Therefore, the tester cannot increase the number of trials (hence test time) without the penalty of fatigue and reduced contraction, which could render the results inconclusive. It seems 200 trials achieves a good compromise with this in mind. Needless to say that if necessary several runs, with breaks in between, could be combined into a grand average with sufficiently low residual noise.