To test the feasibility of using a system for classification of surgical errors and of in-hospital deaths and long hospital stay as markers for errors in surgery we reviewed the hospital records of 273 patients with 285 admissions. During the one year study period there were in all 3767 patients admitted for surgical care. From these we selected the 131 who died in the department during the year, the 100 who had the longest stay (greater than 33 days) and the 91 patients were referred to the departments of internal medicine, infectious diseases or orthopedic surgery. Errors were classified as error of omission or commission, in diagnosis or in therapy. Possible or definitive errors were found in the care of 23% of the patients who died and in 10% of the ones with a long hospital stay. Only 3% of patients referred to other departments experienced errors. It is concluded, that "in-hospital death and "long hospital stay" can be used as markers to identify errors in surgery.