With the world's increasing demand for oil and gas and dwindling onshore reserves, the need to exploit oil and gas has moved into deep water. This move brings with it the potential of accidental releases from well blowouts and pipeline or riser ruptures. While there is a low risk of such accident thanks to today's technology, the oil industry has to be prepared. To better understand how oil and gas would behave during a deep water release, the DeepSpill experiment was conducted in the Norwegian Sea at the Helland Hansen site (65degrees00'N, 04degrees50'E) in 844 m of water roughly 125 km off the coast of central Norway. Four controlled discharges of oil and gas were made during late June 2000 amounting to a total of 120 m 3 of oil and 10,000 standard m 3 of natural gas. The main objectives of the experiments were to calibrate numerical models and to test methods of subsurface surveillance. Extensive observations were made of wind, currents, water density, surface and subsurface oil concentrations, and chemical and biologic samples in the water column. Results showed that the oil started reaching the surface about an hour after the release began and within a few hundred meters of the release site. Oil continued to surface for several hours after the release stopped. No gas hydrates were formed even though thermodynamic equilibrium suggested they should have. No gas bubbles reached the surface indicating that gas dissolution was complete but not as quickly as predicted by standard algorithms. The echo sounders on-board the research vessels were able to track the oil/gas plume as it rose through the water column. In general the surface slick was much thinner than a slick initially released at the surface would have been. Emulsified oil was observed at the surface after the crude oil discharge, with water content increasing with time after the oil came to the surface. An integral plume model [Spill Science and Technology Bulletin 6 (2000) 103] did a reasonable job of predicting the time to surface and the location of the slick though some tuning of the bubble/droplet sizes, gas dissolution rate, and hydrate formation were needed. Finally, the results showed that all gas was dissolved well beneath the surface suggesting that today's safety restrictions governing surface vessel activity could possibly be revised. (C) 2003 Elsevier Ltd. All rights reserved.