<p>This paper examines the impact of varying geotechnical random variables on the assessment of soil settlement improved by stone columns. The soil and stone column properties are considered the random variables. The variables are divided into the Balaam and Booker method. The reliability index for a target settlement is calculated using three reliability analysis methods: First-Order Second Moment, Point Estimate method, and Monte Carlo simulations. The random variables used in the stone column design are assumed to follow either normal or lognormal distributions, and the impact of their correlations is also evaluated. The results show that the reliability index strongly depends on the unit weight of the saturated soils and on the column diameter, and practically very little on the initial void index, for which the variation in the reliability index is less sensitive. Furthermore, the study allowed for the identification of a reduced set of parameters that must be considered random variables in the reliability analysis of stone columns, in addition to conducting sensitivity analysis.</p>