A field experiment was carried out in June 1996 with a subsurface release of air and oil close to the Frigg Field in the North Sea. One of the purposes of this sea trial was to increase the knowledge concerning the behaviour of the oil and gas during a subsurface blowout. This was done by simulating a subsurface blowout at 106 m depth with a realistic gas oil ratio (GOR = 67) and release velocity of the oil/gas. In addition to the main oil release (43 m(3) Troll crude), other releases with water and air (GOR = 7-65) were performed. Important and unique data were collected during these subsurface releases that can be used to improve numerical models for such releases. The main conclusions from these trials are: The field methodology used to study blowout releases in the field appears to be appropriate. The surface oil slick formed by subsurface releases is significantly wider and thinner than if it was caused by a surface release. Only 15-20% of the released oil was detected at the surface. This percentage is expected to depend on factors like depth, GOR, oil type and release velocity. Some discrepancies between existing models and field results were found. The field data may thus serve as a basis for model improvements.