Transmission electron microscopy (TEM) has nanometre resolution and can be used to distinguish single extracellular vesicles (EVs) from non-EV particles. TEM images of EVs are a result of operator image selection. To which extent operator image selection reflects the overall sample quality, and to which extent the images are comparable and reproducible, is unclear. In a first attempt to improve the comparability and reproducibility of TEM to visualise EVs, we compared operator image selection to images taken at predefined locations from the same grids, using four EV TEM preparation protocols, a single EV-containing sample and a single TEM instrument. Operator image selection leads to high-quality images that are more similar between the protocols. In contrast, images taken at predefined locations reveal differences between the protocols, for example in number of EVs per image and background quality. From the evaluated protocols, for only one protocol the operator image selection is comparable to the TEM images taken at predefined locations. Taken together, operator image selection can be used to demonstrate the presence of EVs in a sample, but seem less suitable to demonstrate the quality of a sample. Because images taken at predefined locations reflect the overall quality of the EV-containing sample rather than the presence of EVs alone, this is a first step to improve the comparability and reproducibility of TEM for monitoring the quality of EV-containing samples.