Due to lack of soil sampling during a conventional cone penetration testing, it is necessary to classify soils based on recorded tip and sleeve friction values. Currently available semi-empirical methods of Robertson and Wride (1997) and Olsen and Mitchell (1995) exhibit a significant variability in the estimation of soil type based on cone penetration test (CPT) data. Thus within the confines of this paper it is attempted to present a new probabilistic CPT-based soil classification methodology which addresses the uncertainties intrinsic to the problem. For this purpose, a database composed of 4 10 CPT data pairs of tip resistance (q(c)), friction ratio (R-f) and soil classification based on Unified Soil Classification System (USCS) was compiled. Soil classification was performed by laboratory testing of the disturbed samples retrieved from the boreholes within 2 in of each CPT hole. It is intended to develop a probabilistic model, which will correlate CPT tip and sleeve friction values to actual soil classification.