Theoretical prediction of surface stellar abundances of light elements -- lithium, beryllium, and boron -- represents one of the most interesting open problems in astrophysics. As well known, several measurements of 7-Li abundances in stellar atmospheres point out a disagreement between predictions and observations in different stellar evolutionary phases, rising doubts about the capability of present stellar models to precisely reproduce stellar envelope characteristics. Light elements are burned at relatively low temperatures (from 2 to 5 MK) and thus in the evolutionary stages of a star they are gradually destroyed at different depths of stellar interior, in dependence on the stellar mass. Their surface abundances are strongly influenced by the nuclear cross sections, by the extension of the convective envelope and by the temperature at its bottom, which depend on the characteristics of the star (mass and chemical composition) as well as on the energy transport in the convective stellar envelope. In recent years, a great effort has been made to improve the precision of light element burning cross sections. However, theoretical predictions surface light element abundance are challenging because they are also influenced by the uncertainties in the input physics adopted in the calculations as well as the efficiency of several standard and non-standard physical processes active in young stars (i.e. diffusion, radiative levitation, magnetic fields, rotation). Moreover, it is still not completely clear how much the previous protostellar evolution affects the characteristics of a stellar model and thus the light element depletion. This paper presents the state-of-the-art of theoretical predictions for protostars and PMS stars, discussing the role of several input physics on the stellar evolution.