Recent theoretical work by Shichijo, Hess, and Stillman on a Monte Carlo simulation of high‐field transport and impact ionization in GaAs is examined. The failure of that calculation to reproduce the experimentally well‐documented orientation dependence of impact ionization can be directly related to the use of a phonon scattering rate that is unrealistically high. It is shown that such high scattering rates lead, by the uncertainty principle to a large collisional broadening (?0.3 to 0.6 eV) of the conduction band, thus invalidating the Monte Carlo simulation and rendering questionable any attempt to relate transport to the band structure. The important role played by the avalanche region width in the orientation dependence of impact ionization is also discussed.