Analysis of the settlement at five sites on San Francisco Bay Mud and two sites on Boston Blue Clay has shown that where dessicated crust has developed on the clay, spatial variations in the preconsolidation pressure are the most important cause of differential settlements. A simple probabilistic method is described that can be used to establish confidence limits for settlements at sites with a desiccated crust. Variations in preconsolidation pressure are accounted for by estimating confidence limits for a parameter termed the “radius of preconsolidation.” Application of the methodology to the settlement of fills on San Francisco Bay Mud and Boston Blue Clay has shown that the method produces results in good agreement with field measurements and that it is suitable for use in practice. The method can be used for clay deposits which have an overconsolidated layer and requires no more testing than is performed for conventional settlement analyses. However, unlike conventional settlement analyses, it provides an estimate of the likeli-hood that the settlements will vary from the mean by a given amount due to spatial variations in soil properties.