Protection of vulnerable marine ecosystems (VME) is a critical goal for marine conservation. Yet, in many deep-sea settings, where quantitative data are typically sparse, it is challenging to correctly identify the location and size of VMEs. Here we assess the sensitivity of a method to identify coral reef VMEs based on bottom cover and abundance of the stony coral Solenosmilia variabilis on deep seamounts, using image data from a 2018 large survey off Tasmania, Australia. Whilst there was some detectable influence of varying coral cover and the abundance of live coral heads, the distribution of coral reef VMEs was not substantially shifted by changing these criteria or altering the attributes of a moving window used to spatially aggregate coral patches. Whilst applying stricter criteria for classifying VMEs predictably produced smaller areas of coral reef VME, these differences are not sizeable and were often negligible. A model prediction of the area of suitable habitat for coral reef in the Tasmanian area was much greater than that estimated in this study. Coral reef VMEs formed large contiguous ‘blankets’, mainly on the peaks and flanks of seamounts, but were absent from the continental slope where S. variabilis occurred at low abundance (cover) and/or had no living colonies. The true size of the Tasmanian coral reef VMEs ranged from 0.02 to 1.16 km2; this was relatively large compared to reefs of S. variabilis mapped on New Zealand seamounts, but is small compared to the scales used for regional model predictions of suitable habitat (typically 1 km2 grid cell), much smaller than the smallest units of management interest (100s – 1000s km²). That the method is not overly sensitive to the choice of criteria is highly encouraging in the context of designing spatial conservation measures that are robust, although its broader application, including to other VME indicator taxa, needs to be substantiated by scenario testing in different environments. Importantly, it should encourage uptake by the broader community (e.g. fishing and mining sectors) and forms the basis for better predictive VME models at larger spatial scales and beyond single taxa.