A stochastic modeling procedure, designed to capture sequence stratigraphic principles, has been developed for modeling fluvial reservoirs where high-frequency base-level fluctuations have exerted a strong influence on reservoir architecture. Reservoir stratigraphy and architecture are defined by the successive simulation of two types of surfaces: base-level rise (flooding) surfaces and base-le-c el fall surfaces (sequence boundaries). The flooding surfaces are modeled as standard two-dimensional Gaussian fields. Modeling sequence boundaries with incised valleys is more complex and required the development of a novel :"object" modeling technique. This new model call be used to generate realistic valley geometrics and is flexible enough to allow for complex multiwell conditioning. The modeling procedure is illustrated using a test data set based on well interpretations from the fluvial Statfjord Formation in the Statfjord field of the Norwegian North Sea. The main reservoir sandstones were deposited in valleys defined by a sequence boundary at the base and a flooding surface at the top, whereas the main barriers to flow are mudstone-rich intervals deposited on unconfined alluvial plains, Five sequence boundaries and four flooding surfaces have been interpreted within an approximately 60-m-thick reservoir interval, Simulation of these surfaces using the new modeling procedure defines the three-dimensional distribution of reservoir units and barriers. The modeling procedure allows the simulation of realistic geometries that are in accordance with the geologist＇s conceptual model for the reservoir The models also provide an enhanced description of reservoir distribution and connectivity, and can serve as an improved basis for reservoir management, well placement, and predictions of reservoir performance in complex fluvial reservoirs.