INTRODUCTION The Troll field is a huge oil and gas field situated in the North Sea approximately 100 km northwest of the city Bergen of Norway (Figure 1). The oil leg in the Troll West oil province (TWOP, 22-28 m thick) and the Troll West gas province (TWGP, 12-14 m thick) is predicted to contain about 600 x 10 6 Sm 3 of oil in place; however, the combination of a layered structure with 2-6° dip and a relatively thin oil leg makes the oil prospects very uncertain with respect to optimal target locations. Only minor vertical adjustments of the highly permeable sands may lead to significant changes in the lateral position of these prospects. In this paper, we present the stochastic modeling study (SMS) on Troll West in 1992, comprising mainly structural modeling, and describe the principles and assumptions related to the modeling. During the model building, we tried, as far as possible, to follow the same concepts as in deterministic modeling. The main structural control of Troll West yields the seismic Top Sognefjord reflector (top reservoir) and a more blurred Top Fensfjord reflector (base reservoir). Recently an intrareservoir reflector, 4.1TR, corresponding to a near-top 3C sand, was mapped in TWOP and parts of TWGP. This framework, which is fundamental for the location of individual reservoir layers in undrilled areas, involves uncertainties related both to the seismic interpretation and processing, as well as to the depth conversion method. The subseismic reservoir layering was sampled only at well locations. Due to this spatial undersampling, the uncertainties may be severe.