Several statistical estimators that may be used in neutron Monte Carlo calculations are compared by means of idealized problems. The objective is to evaluate these estimators for use in the study of neutron thermalization and thermal neutron behavior by Monte Carlo techniques. It is concluded that the minimum variance linear combination of the analog estimator and the modified path length estimator (proposed recently by Spanier) is the best of these estimators  for such applications. Where computer limitations require the use of a simple (rather than a combination) estimator, the modified path length estimator is recommended.