The advent of digital pathology and the consequent use of virtual microscopy have allowed insights into the processes used by pathologists to interpret a histopathological slide. Recently Roa-Pena, Gomez, and Romero[1] have added to the literature by studying the navigation strategies of four pathologists reading a very small case set of six virtual slides depicting different organ/disease combinations. The pathologists’ task was to identify the organ and, whenever possible, to provide a diagnosis of the disease process present. They used a custom-made interface, with two images depicted side-by-side in a single display: (1) a large thumb image showing the entire virtual slide, where all navigation, panning, and zooming activities were carried out by moving and/or modifying a small ‘view’ window; and (2) a large resolution image of the ‘view’ window. The authors reported that about half of the time spent reading the slide was used to closely examine areas of interest, whereas the other half was spent navigating through the piece, zooming, changing the ‘view’ window, etc. This is in agreement with a ‘search-and-focus’ strategy that has been reported elsewhere.[2] However, a unique aspect of this study is that the authors used different criteria to determine the pathologists’ ‘regions of interest’ (ROI). Among them (i) a group-based ‘coincidence’ criterion, according to which a given area had to be examined by more than one pathologist; and (ii) an ‘individual’ criterion, which took into account any areas that a single pathologist spent some time examining. Using these criteria, they reported a high ‘coincidence’ rate of areas visited by more than one pathologist, ranging from 41% in a gallbladder sample to 97% in a sample of the endomyometrium (mean: 70.5%). However, when the authors contrasted the ROIs that were determined using the ‘coincidence’ and the ‘individual’ criteria, their results suggested a low level of agreement. This led them to conclude that individual pathologists search each slide using their own unique strategies, and no general patterns can be observed despite the high ‘coincidence’ rates. While these results are somewhat surprising, before taking them at face value one needs to consider the very small sample size of the study and the custom-made interface with its own unique characteristics. I will address these results in the following commentary.