The classical Broido-Shafizadeh model describes cellulose pyrolysis as two competing reactions with high activation energies, E tar and E gas . The reactions with E tar and E gas lead to volatile tar (predominantly levoglucosan) and to light gases together with char, respectively. The equations for peak temperatures and yields of any competing reactions prove the following fundamental rule. With increasing temperature, a pathway with higher activation energy suppresses a pathway with lower one. The original hypothesis explains the channel with E gas by existence of the so-called ''anhydrocellulose'' arising due to the low-temperature cross-linking of cellulose by dehydration. If it is so, E tar > E gas . However, the direct analysis of pyrolytic gases shows that, in fact, the gasification is the high-temperature pathway, thus, E gas > E tar . Disregarding real data about gasification, the Broido-Shafizadeh model erroneously fastens the channel ''gas'' to the char yield decreasing with increasing temperature. This leads to the contradiction. The alternative model ignoring the notion of ''anhydrocellulose'' was proposed for getting rid of it. The model interprets mass loss by two competing pathways of cellulose depolymerization, namely, by transglycosylation with E tar % 190-200 kJ/mol and by E ielimination with E gas % 250 kJ/mol. Derivatives of glucose arising due to the E i -elimination convert into char and light gases. The secondary competition between charring and gasification explains all observable effects.