The original STOIIP of the Fulmar Field is currently estimated as 853 x 10 6 STB and it is considered that approximately 540 x 10 6 STB of oil has so far been produced, representing 63.4% of the original reserves. This high recovery factor coupled with an increasingly high water-cut (currently averaging 90%) suggests that the end of field life is drawing close. Prior to making an abandonment decision however it was considered prudent to review and update both the geological and dynamic reservoir models in order to ensure that the full field potential has been realized before making provisions for abandonment. In this context, the depositional model and stratigraphic architecture of the Upper Jurassic Fulmar Sands have been reviewed. Two geological models have been considered: (1) the preexisting model; and (2) an alternative model, which reflects continuing uncertainties in the field's stratigraphic architecture.The main part of the Fulmar reservoir consists of a stacked sequence of shoreface sediments that were deposited within an extensional, back-stepping shallow marine setting. This sequence is punctuated by several minor flooding events, which are characterized by distal, poorer quality shoreface sediments. The latter deposits locally form small, but distinct, pressure and flow discontinuities in the reservoir. These heterogeneities were captured in detail, in a 3D geological model and input into the full field reservoir simulation using the Shell proprietary GEOCAP/MoReS software. The history match of the dynamic reservoir model predicted that by-passed oil could be locally trapped below these flooding events. This concept was confirmed in mid 1997 by two wells drilled on the crest of the field. The simulation model also indicates that a significant volume of potentially unswept oil may be present on the NE flank of Fulmar within the poorer reservoir quality, lower shoreface 'Clyde sands'. This concept will be tested by drilling a horizontal well into these sands in the mid part of 1998. As the architecture of the alternative geological model showed some differences to the pre-existing field geological model, particularly in the down flank areas, this model was also input into the GEOCAP simulation in order to compare and test the reliability of the simulation predictions. Only minimal differences were observed between the two simulation models, which reflects the dominantly high net: gross (N: G) in the main producing western part of the field and the similarity of the correlations in the crestal and eastern parts of the field.Both similation models show a significant improvement in the history match, which highlights the potential benefits to be gained by constructing detailed 3D reservoir models that better describe the architecture of the flow units and barriers in the reservoir than more 'traditional' layer-based (CPS-3) models. These studies are helping to maximize recovery from this highquality, strongly depleted reservoir.This paper primarily presents a review of the stratigraphic architecture model...