The sampling distribution of Bovicola ovis (Schrank) on sheep was examined in two flocks, one with a light and one with a heavy infestation of lice. The derived distributions were used to calculate the sensitivity of detecting lice on individual sheep and in flocks by fleece parting regimes that varied in number of parts per animal and number of sheep per flock, different scenarios of flock sizes, proportion infested and louse density were examined. Lice were aggregated among fleece partings in the heavily infested flock and described by a negative binomial distribution with k values between 0.3 and 1.92. The distribution was indistinguishable from Poisson in the lightly infested flock. The assumed distribution had little effect on sensitivity, except when only one fleece part per animal was examined. On individual sheep where louse density was 0.5 per 10 cm part or greater, there were only marginal gains from inspecting more than 10 parts per animal. Increasing the number of sheep inspected always increased sensitivity more than increasing number of parts per sheep by an equivalent amount. This advantage was greatest in situations where a low proportion of sheep in the flock were infested with a high density of lice, and less where a low proportion of sheep were infested with a low density of lice, or a high proportion of sheep were infested with a high density of lice.