Up to now, it has been very challenging to directly observe defect creation by blistering. An even more difficult task is to identify criteria for the actual nucleation of blisters. We recently succeeded to create intragranular blisters in the transparent region of tungsten samples thinned for transmission electron microscopy. From a narrow zone without blisters near the central perforation of the sample, we were able to deduce a minimum W thickness of about 50 nm that is necessary for blister nucleation. We also observed a massive increase of the dislocation density due to blistering of about a factor 100 compared with the initial material. These dislocations predominantly exist close to observable cavities. A unique feature of creating blisters in electron-transparent samples is that this provides the opportunity to correlate the occurrence of blisters with pre-existing defects. In addition to observing blisters, we were also able to study the DSSL created during plasma exposure by TEM and investigate its defect structure.