Percutaneous gastrostomy tubes which have non-deflatable “anchors” in the stomach can be removed by several methods, including cutting the tube, pulling it out using traction, or removing the anchor under direct endoscopic visualization. Because of the risks and discomfort involved with the latter two methods, cutting the tube has emerged as a more appealing option. However, complications may arise if the anchor does not pass through the gastrointestinal tract. We have therefore reviewed our experience with gastrotomy removal over the past 4 years. From January 1993 - June 1997, 44 percutaneous endoscopic gastrostomies (Bard™) were inserted. 17 patients have their original tubes, 5 tubes were removed under direct endoscopic guidance, 6 were pulled out, and 16 have been cut. Of these 16, 6 did not pass through the GI tract, including 3 15 Fr. tubes and 3 20 Fr. tubes. 2 of the 20 Fr. anchors became lodged in the esophagus. In one patient, vomiting was noted within 24 hours, and in the other, mild respiratory distress was noted after 2 weeks of intermittent vomiting. The 4 other patients were asymptomatic and their anchors (3 15 Fr., 1 20 Fr.) were removed from the stomach endoscopically without complications. The 6 patients who did not pass the anchor averaged 1.62 +/- 0.72 years old. Patients who did pass the anchor averaged 6.25 +/- 5.25 years of age. The mean weight was 9.2 +/- 2.8 kg in the group that did not pass the anchor versus 21.95 +/- 14.2 kg in the other group. The underlying diagnosis was unrelated to passage of the anchor. 9/10 patients who passed the anchor were contacted to determine when the anchor passed. 2 patients never saw the anchor in the stool but had normal X-rays. The remaining 7 recalled passage of the anchor at a mean of 15 days (range 12-24). We conclude that over one third of patients whose gastrostomies are cut may not pass the anchor spontaneously. Of patients that do, 2-3 weeks may be required for spontaneous passage. The primary risk factors for not passing the anchor were low age and weight, whereas underlying diagnosis or large tube size did not increase the risk.