Accurate, repeatable estimates of population densities are often desired for vegetation monitoring. However, conventional transect and quadrat field sampling methods are not always applicable to plants such as the rare shrub Epacris muelleri Sond., growing on largely inaccessible cliffs and rock faces. E. muelleri is an ericaceous shrub restricted to the Blue Mountains region in New South Wales, which is to be monitored to detect potential effects of underground coal mining. In this manuscript, we evaluated observer error associated with density estimates to assess suitability of applying the timed-meander method to this species. The results indicate that a visual search method using binoculars can generate repeatable results among observers with very different experience levels. However, there is a large margin of error in estimating density when there are many plants growing in close association or overlapping on a cliff. Nonetheless, with a strict set of protocols and further evaluation, this method shows promise as a rapid yet robust method for carrying out repeatable surveys for quantifying changes in the population.