Determining the centrality of a collision, or the impact parameter, in heavy-nucleus collisions is important for the analysis of all experimental data and their comparison with theory. One of the methods for this is to measure the number of spectators using a hadron calorimeter arranged at a small angle to the nuclear beam. It is shown that the accuracy in determining the impact parameter at the attainable energy resolution of the hadron calorimeter is inadequate for using the calorimeter in the MPD/NICA and CBM/FAIR projects. The error is as large as 35% at the beam energy of 2.5 GeV even for peripheral collisions. Secondary processes during the motion of spectators through a nucleus make an additional contribution to the error for central and mid-central collisions.