The 'New Deal' set out by the Department of Health in 1991, together with the introduction of specialist 6-year training grades by Calman in 1993, has resulted in a decrease in available training time for surgeons in the UK. There is also an emerging belief that surgical procedures performed by trainees might compromise patient outcome. This study examines the level of trainee experience in a specialist gastrointestinal unit and whether operation by a trainee surgeon adversely affects patient outcome. All patients in the University Department of Surgery, Royal Infirmary, Edinburgh, undergoing oesophagogastric, hepatic or pancreatic resection between January 1994 and December 1996 were entered into the study. The early clinical outcome (in-hospital mortality and morbidity, considered in three groups: anastomotic leak, other technique-related complications and non-technique-related complications) was evaluated with regard to the grade of surgeon (consultant or trainee) performing the operation. Of the 222 patients undergoing major upper gastrointestinal resection during the study period, 100 (45%) were operated on by trainees. Trainees were assisted and closely supervised by consultants in all but six resections. There was no major difference in mortality rate (consultant, 4.1% vs trainee, 5%), incidence of non-technique-related complications (consultant, 6.7% vs trainee, 7.1%), anastomotic leaks (consultant, 10.7% vs trainee, 5%) or technique-related complications (consultant, 18.9% vs trainee, 15%) between the two grades of surgeon. In a specialist unit, the early clinical outcome of patients undergoing major upper gastrointestinal resection by supervised trainees is no worse than in those operated on by consultants. Participation of trainees in such complex procedures enhances surgical training and does not jeopardise patient care.