Résumé: Traditionally, the design of geotechnical structures is based on a deterministic approach in which all parameters take a fixed value, which leads to an oversized and unjustified underestimation of the bearing capacity of soil, as well as the overestimation of stress. However, the effects on structure safety of uncertainties associated with the design parameters are not quantifiable. An alternative method with which to study the reliability of geotechnical structures is based on the theory of probability and involves the application of partial safety factors for all design parameters (random variables). These factors are derived using probabilistic methods and take into account the dispersion of soil parameters (stochastic model). In this paper, a benchmark for a strip footing with axial load was used, with security expressed via a probability of failure or reliability index and evaluated by means of a universal computer code based on probabilistic methods. Analysis is carried out by considering the various types of parameter distributions, thereby enabling a better assessment of the effects of uncertainty and the identification of a set of parameters with high incidence.