Abstract. We study the question of what dierence it makes for the derived ®eld-aligned conductance (u) values if one uses Maxwellian or kappa distributions for the ®tting of low-orbiting satellite electron¯ux spectra in the auroral region. This question has arisen because sometimes a high-energy tail is seen in the spectra. In principle, the kappa ®ts should always be better, because the kappa distribution is a generalization of the Maxwellian. However, the physical meaning of the parameters appearing in the Maxwellian is clearer. It therefore makes sense to study under which circumstances it is appropriate to use a Maxwellian. We use Freja electron data (TESP and MATE) from two events. One of the events represents quiet magnetospheric conditions (stable arc) and the other represents disturbed conditions (surge). In these Freja events, at least, using kappa rather than Maxwellian ®tting gives a better ®t to the observed distribution, but the dierence in u values is not large (usually less than 20%). The dierence can be of either sign. However, sometimes even the kappa distribution does not provide a good ®t, and one needs a more complicated distribution such as two Maxwellians. We investigate the relative contributions of the two Maxwellians to the total ®eld-aligned conductance value in these cases. We ®nd that the contribution of the highenergy population is insigni®cant (usually much less than 20%). This is because u is proportional to na i p , where n is the source plasma density and i is the characteristic energy.