This paper presents a characterization of the variability of a lightly overconsolidated and highly sensitive clay deposit located near Saint-Hilaire, 50 km east of Montreal. The geotechnical investigation consisted of in situ and laboratory tests. The variability of the in situ test results is the subject of this paper. The working hypothesis assumes that piezocone and vane test results may be modelled by a random function. This is done on the basis of a geostatistical approach. In situ vane and piezocone tests are found to increase with depth following a linear trend. This is a nonstationary problem and inference of the autocorrelation function must be made through the estimation of a generalized covariance. Results for both types of tests give the same shape of generalized covariance. Measurements made with both testing devices yield the same 2 m autocorrelation distance but the standard deviations are different. The standard deviations for the piezocone cone resistance, pore pressure behind the cone ti...