Percutaneous endoscopic gastrostomy (PEG) could be potentially more hazardous in patients with prior abdominal surgery. To minimize risk of complications following PEG, we have modified the conventional method to include selection of a "safe gastrocutaneous fistulous tract," using an aspirating, lidocaine-filled syringe and needle. "Safe tracts" prior to PEG placement were determined by simultaneous air return in the aspirating syringe and endoscopic visualization of the intragastric needle. This technique was used to compare the results of PEG in 27 operated and 80 nonoperated patients. Our results show that morbidity and mortality rates and types of complications following PEG in previously operated and nonoperated patients are comparable, but technical success rates are slightly lower in the surgical group (88 vs 100%). The incidence of failed PEGs was highest in patients with partial gastric resection compared to those with intact stomachs (29 vs 5.0% p less than 0.05). PEG was safely and successfully performed in all cases (surgical and nonsurgical) when safe tracts were confirmed. Selection of a safe tract should be made prior to PEG placement, especially in patients with prior abdominal surgery.