In common practice, when designing a geotechnical model of a landslide, some stratigraphic details are usually discarded in order to simplify the problem. This approach is supported by the hypothesis that the significance of the results of the model would not be affected, besides additional uncertainty associated with the thickness and lateral extent of these minor strata is avoided. In this work a new method called Boolean Stochastic Generation (BoSG) which relies on a Monte Carlo generation of numerous soil layers distributions following a Boolean logic (the material is either matrix or layer) is associated with the Point Estimate Method. This association allows to determine the influence of the variation of the soil parameters of the matrix material in the BoSG runs. The Mortisa landslide was selected as case study as it shows a soil composition particularly suitable for BoSG: gravel lenses interdigitated in a silty-clay matrix. Results show that the bigger is the difference between matrix and layer properties, the greater is the influence of more resistant layers as they are intercepted by the plastic deformation. Moreover, studying the distribution of the uncertainty along the slope may help to indicate where to perform a secondary investigation campaign.