This paper shows how bandgap EEL spectra are commonly processed either by d
econvolution or subtraction methods in an attempt to remove effects arising
from the finite width and long tail of the zero loss peak. This paper will
compare the two main methods, and show that the deconvolution method is si
gnificantly more reliable and free of user-interpretation or artefacts. We
first consider how the zero loss peak is produced and what effects experime
ntal conditions have upon it, and then show that only by the deconvolution
method can accurate bandgap energies and bandgap densities of states be det
ermined.