We have calculated the Luttinger-like parameters, used in the 8 x 8 Hamilto
nian describing both conduction and valence bands, via a second-order pertu
rbation theory that allows one to make the Lowdin renormalization valid for
a set of nondegenerate energy levels. We show that the formulas given in t
he paper by Pidgeon and Brown [Phys. Rev. 146, 575 (1966)] are the correct
ones, contrary to what is claimed in recent papers. This leads to strong di
screpancies when the spin-orbit splitting is not much smaller than the band
gap. For example, the ratio between the Luttinger-like parameters <(<gamma
>)over tilde>(1), calculated in the two cases, is of the order of 1.5 in Ga
As, 3 in InAs, and 16 in InSb. The accuracy of the equations used by Pidgeo
n and Brown is also discussed.