We studied 202 patients admitted to two major teaching hospitals for planned gastrointestinal surgery to assess the ability of several techniques of nutritional assessment to predict major postoperative complications (infection and/or wound problems). Subjective global assessment (SGA) and albumin were both of predictive value, and combinations of these variables were useful in differentiating low-risk from high-risk patients. Transferrin, creatinine-height index, percent ideal weight, percent body fat, and total lymphocyte count were not useful in predicting complications. We conclude that SGA and albumin are useful "nutritional assessment techniques" for patients undergoing major gastrointestinal surgery if the purpose of such an assessment is to predict postoperative "nutrition-associated complications." The second major finding of this study was the unexpectedly low rate of complications (10%) which was found in both hospitals. We suggest that these low complication rates may be more generalizable to patient populations derived from a wide community base, rather than those described in other studies in which veterans or patients of lower socioeconomic status comprised the sample.