We discuss the procedure of calculating the density of states in the gap for amorphous silicon by integrating the excess sub‐band‐gap absorption. We show that in general the proportionality coefficient between the density of states and the integrated sub‐band‐gap absorption is not unique. The sum rule has been used improperly since integration of the excess subgap absorption is terminated either at a fixed energy or at an energy which does not include all optical transitions to the conduction band. We estimate the errors that can arise from this procedure.