This study provides the theoretical background of the decision to count approximately 750-1,300 neurons per individual in the preceding study of Heinsen et al. [6] finding a significant (P < 0.05) nerve cell loss in the thalamic mediodorsal nucleus in Huntington's disease with the so-called V(Ref) x N(V) method. Using a computer simulation of the study of Heinsen et al., it was shown that the legitimation for counting only 100-200 neurons per individual in previous studies comparable to that carried out by Heinsen et al. was based on incorrect assumptions. In this context it was of particular importance to confirm the theoretical prediction in the literature that the random error of total neuron number estimates obtained with the V(Ref) x N(V) method is actually greater than assumed in current stereological studies. In summary, this study revives the question of how many individuals need to be investigated and how many neurons (or other cell types, respectively) need to be counted per individual in studies comparable to that carried out by Heinsen et al.