At various sites on three grids with crude-sap preparations of tobacco mosaic virus (TMV), electron micrographs were made and per photograph 50–200 particles were measured. Calculated average lengths turned out to differ up to 4% according to the site photographed. Calculated lengths appeared to depend on potentiometer settings required for focussing. This is due to the effect of preparation height or object position along the optical axis of the electron microscope on image magnification. This is further supported by the effect of additional rings underneath the grid and of artificial bending of the grid. Because it is hard to reproduce exactly electron microscope magnification, and at given potentiometer settings per photograph calculated average lengths may still vary up to c. 4%, the application of internal standards is essential for determining the size of unknown viruses. With TMV only relative particle sizes can be determined, and even with more accurate non-biological standards no absolute virus sizes will be found.