[en] The straightforward method for calculation of resonance self-shielding is to generate one or several resonance ladders, and to process them as resolved resonances. The main drawback of Monte Carlo methods used to generate the ladders, is the difficulty of reducing the dispersion of data and results. Several methods are examined, and it is shown how one (a regularized sampling method) improves the accuracy. Analytical methods to compute the effective cross-section have recently appeared: they are basically exempt from dispersion, but are inevitably approximate. The accuracy of the most sophisticated one is checked. There is a neutron energy range which is improperly considered as statistical. An examination is presented of what happens when it is treated as statistical, and how it is possible to improve the accuracy of calculations in this range. To illustrate the results calculations have been performed in a simple case: nucleus ²³⁸U, at 300 K, between 4250 and 4750 eV. (author)