[en] We study the problem to accurately account for the heavy quark threshold effects in hard processes. We employ the direction perturbative Feynman diagram analysis and the Stueckelberg-Bogoliubov massive renormalization group formalism to show that both methods results in the same prescription for the 'physical' mass-dependent QCD effective coupling as an argument of anomalous dimensions that determine the evolution of structure function (parton distributions). By considering the one-loop example, we demonstrate that an intrinsic ambiguity of the standard approach based on the notion of 'effective number of quark flavours' may affect the α_s determination at the level of two-loop effects