Abstract Statoil has been involved in exploration in the Gulf of Thailand since 1985, following an invitation by PTTEP (National Exploration and Production Oil Company of Thailand) to assist in the evaluation and possible development of the Bongkot gas and condensate Field. Statoil and PTTEP jointly studied the field in 1985 and 1986. In 1990 PTTEP farmed 60% out to a group consisting of Total, British Gas and Statoil. Development work on the field started in 1990 with the acquisition of 3D seismic data. The field came on stream in June 1993, and production increased from the original planned 150 10 6 ft 2 /day to 350 10 6 ft 2 /day at present. In the joint PTTEP/Statoil study the Oligocene to Middle Miocene reservoirs were described as fluvial point bar sandstones, deposited in a meandering river system located on a delta top or coastal plain. Based on available data it was not possible to map the individual reservoirs. 22 wells found intermingled gas- and water-bearing sandstones without obvious relation to structure. No field fluid contacts were found, and the pressure regime was thought to be hydrostatic. Statoil simulated a reservoir model consisting of numerous pointbar deposits on a muddy delta plain. The pointbars were assumed to be isolated sandbodies which favoured stratigraphic trapping. Structural mapping of existing 2D and old 3D data sets was thought to give reliable structural maps, although it was difficult to correlate single events. Stratigraphic traps were interpreted as the main traps in the field because no relationship was found between structural closure and gas pools. Since 1990 when field development started with Total as operator around 2500 km 2 of 3D data has been acquired, and approximately 20 exploration wells and 45 development wells have been drilled. These data resulted in a much more accurate description of the field, especially through the use of high-resolution 3D seismic data. The well information allowed the identification of high GR intervals with a continuous seismic expression. Also individual sand reservoirs could be correlated. It became evident from the detailed 3D seismic interpretation that structural traps form the main trapping mechanism, although a minor part of the reserves are found in isolated channel sands, which could be detected from amplitude maps. It has been difficult to identify stratigraphic traps but it is hoped that ongoing seismic attribute analyses will help. In 1993 Statoil farmed in to licence B10/32, operated by Ampolex. The Bongkot model of isolated fluvial channels, with traps formed by several faults with or without structural closure was accepted as an attractive play type. Not withstanding this, the first wildcat well, Karawake-1 was a structural test. The well failed to find hydrocarbons because it intersected a series of overlapping alluvial fans; these lack internal seal and would have precluded hydrocarbon migration and entrapment. Ampolex made a reinterpretation of the block based on the existing 2D seismic, and concentrated on seismic facies mapping. The mapping resulted in two prospect areas where fluvial facies could be recognized above a lacustrine source rock; and also the results of the Karawake-1 well could be better explained. A second well should prove the model.