The size of the putative export channel in the bacterial flagellar filament appears small (25 A) in studies done by electron microscopy but large (60 A) in studies done by X-ray diffraction. We have undertaken additional studies by electron microscopy to examine some of the possible causes of the difference. A comparison of three-dimensional image reconstructions of native and reconstituted filaments rules out the presence or absence of flagellin monomers in the export channel as the source of the variation in apparent channel size. The channel seen in reconstructions from both kinds of filaments is 25 A in diameter. The difference in the previous studies is more probably a result of artifacts introduced in either the X-ray or the electron microscopical methodology. Comparisons of three-dimensional reconstructions from images of filaments embedded in various stains (anionic, cationic and neutral) and in ice, taken at a range of defocuses, rule out the two most likely sources of artifact in electron microscopy (i.e., staining artifacts and defocus phase contrast). Based on these studies we suggest that the channel seen in the image reconstructions is free of exported flagellin monomers, that its true diameter is about 25 A, and, therefore, that the flagellin monomer must be unfolded to pass along it.