The evolution of QCD coupling constant at finite temperature is consid
ered by making use of the finite temperature renormalization group equ
ation up to the one-loop order in the background field method with the
Feynman gauge and the imaginary time formalism. The results are compa
red with the ones obtained in the literature. We point out, in particu
lar, the origin of the discrepancies between different calculations, s
uch as the choice of gauge, the breakdown of Lorentz invariance, imagi
nary versus real time formalism and the applicability of the Ward iden
tities at finite temperature.