Geotechnical performance of 'soil-structure' systems is strongly dependent on the properties of the soil and prediction of the performance of these systems in real conditions requires accurate modeling of soil parameters. With the help of high-speed computers, now it is possible to create advanced constitutive models, but large uncertainties and variations in soil properties could reduce the advantages gained by using such models. In this paper sources and types of uncertainty in geotechnical engineering practice are first presented, followed by a review of the basic concepts and terminology of geostatistics. Finally, procedures for quantification of uncertainty and for geostatistical estimation and simulation of spatially variable soil properties are presented.