The Gullfaks Sor oil- and gas-filled Brent reservoir can be divided into 14 fault segments, five of which are penetrated by wells that have proven hydrocarbon columns between 32 and 316 m. The uncertainties in gas and oil in-place volumes have delayed the time from discovery in 1978 to the current partial development plan that was submitted to the Norwegian authorities in 1995 for development of the oil reserves. This paper describes the method used for modeling fault-segment reservoir heterogeneities and the method of quantifying the distribution of oil and gas accumulations. The final product is the most likely stock tank oil initially in place/gas initially in place (STOIIP/GIIP) with reliable high-and low-volume estimates for management decision making. We evaluated the initial reservoir fluid pressures and fault-sealing properties together with observed internal reservoir barriers to vertical fluid flow and several fluid-contact observations. We established a quantification of the oil-leg thicknesses, including uncertainty ranges, on the basis of the field gas-and water-pressure gradients and the fluid densities. This quantification had to honor the various fluid contacts, the fluid gradients, and log observations, such as water up to, gas down to, and geochemical core analysis. We evaluated the degree of communication between drilled fault segments by determining probabilities of static-pressure communication between the drilled and undrilled fault segments and between the various reservoir units across the faults. We assessed communication of fluid flow across faults by evaluating the clay smear (amount of clay material in relation to the fault planes) and juxtaposition relations of the reservoir units across faults. We used a statistical approach to quantify the uncertainty in the proven and probable in-place Gullfaks Sor Brent formation hydrocarbon volumes for field-development decision making. A geostatistical program designed to handle laterally varying uncertainties was used for quantifying uncertainties in the seismic interpretation, velocity model, depth conversion, isochores, and fluid contacts. To evaluate the petrophysical uncertainties and the probabilities for gas-or oil-filled reservoir segments, we used a classical Monte Carlo simulation method. The combined use of geostatistical analysis and Monte Carlo simulation has proved to be a useful and fast approach for modeling the Gullfaks Sor field.