OBJECTIVE Percutaneous endoscopic gastrostomy (PEG) is now easily used in the event of long-term enteral nutrition. Tolerance of long-term enteral feeding has been documented in different populations but the documentation is incomplete in the case of older and frail people. Our aim was to describe early and late tolerance in this population, and to propose ways in which it could be improved. METHODS Retrospective study in two parts: tolerance and quality of care; case-control study for tolerance. METHODS A nursing home with 240 beds in south-west France. who had undergone PEG insertion between January 1990 and June 1994. Fifty-eight patients were involved, 12 under 65 years (48 +/- 10.6), and 46 over 65 (80.7 +/- 9.3). The gastrostomy insertion was performed because of a vegetative state in 6 patients, swallowing difficulties in 31 and anorexia in 21. A control group was gathered in December 1996, which included all patients for whom the question of nutritional support was mentioned books but where no artificial nutrition had been implemented due to the patients' or families' refusal or to a staff decision. This group included 50 patients, 5 younger than 65 years (54 +/- 8.3), and 45 older (84.7 +/- 7.6). In 22 cases the nutritional problem was swallowing difficulties and in 28 cases anorexia. Pressure ulcers were present before insertion in 34 patients in the PEG group and in 7 of the control group (p < 0.001). METHODS Prognosis, early and late cutaneous digestive (ileus, vomiting, gastroesophageal reflux) and pulmonary (bronchorrhea, dyspnea and aspiration pneumonia) complications for PEG and control groups, and patients with signs of poor behavioural tolerance of PEG were recorded in the chart. Audit of quality of care was performed in the PEG group using eight criteria: two concerned the pre-insertion period, two the early follow-up and four the long-term follow-up. RESULTS Early mortality (4 weeks) was 13.8% in PEG (vs 10%, NS), mid-term mortality (between 4 and 8 weeks) was 12.1% (vs 14%, NS) and late mortality was 19.0% (vs 42.0%, length of follow-up 63.4 +/- 42.1 weeks compared to 53.1 +/- 63.8 weeks, NS). The duration of follow-up of the living patients was 71.6 +/- 61.8 weeks in PEG compared to 48.0 +/- 70.5 in the control group. Only 20% in PEG were free of any cutaneous complication around the insertion site, and 8 abscesses occurred during the first week. Pulmonary complications occurred in 39% of the PEG group (vs 30.0, NS). Aspiration pneumonia was significantly associated with swallowing difficulties in both groups (p < 0.05). Vomiting occurred for 15.5% of the PEG group (vs 12%, NS), ileus in 13.8% (vs 6%, NS). Gastroesophageal reflux was found in 2 PEG patients, compared to 1 case among the control patients. Pressures sores were healing in 20 out of 34 patients in the PEG group (vs 2 out of 7) and new ulcers appeared in 6 out of 24 (vs 8 out of 43). Fifteen (25.8%) of the PEG patients attempted to withdraw the tube.