We retrospectively analyzed a cohort of 178 patients aged 65 and older who received percutaneous endoscopic gastrostomy (PEG) at St. Luke's International Hospital from January 1998 to December 2002. Of all patients, 1.7% (n=3) died within 7 days after PEG placement, the 30-day mortality was 5.9% (n=10), and the 1-year survival was 61.4%. Using single variable analyses, the survival rate was significantly better in groups whose score of serum albumin, total-cholesterol or cholinesterase at PEG placement was high. Using the Cox proportional hazards model, cancer and serum total-cholesterol score were identified as independent predictors of the prognosis. The improvement of nutritional status after PEG placement was significant, but was limited. PEG placement must be decided only when the prognosis and risks estimated on the patient's condition were understood. It is important to consider a good balance between the patient's long-term survival and improvement of QOL.