AbstractTwo common methods of applying dead‐time corrections in EDS systems, both of which are carried out internally in the electronics, are examined. The Barnhart method shows serious discrepancies when making dead‐time corrections and in certain cases no correction is made at all. The problem is associated with the reliance on the fast amplifier to count all x‐rays collected by the detector. In practice, however, the fast amplifier does not always succeed in carrying out this function since x‐rays with energies<1500 eV may fall below its threshold level. The problems become worse when a formvar window is substituted for beryllium since a greater proportion of low energy x‐rays can now reach the detector. The magnitude of the discrepancy is shown also to be a function of specimen and electron accelerating voltage. Similar experiments with an EDS system using the Covell method show that no such discrepanc