OBJECTIVE--To compare percutaneous endoscopic gastrostomy and nasogastric tube feeding in patients with persisting neurological dysphagia. DESIGN--Randomised 28 day study of inpatients requiring long term enteral nutrition. SETTING--Three Glasgow teaching hospitals. SUBJECTS--40 patients with dysphagia for at least four weeks secondary to neurological disorders: 20 patients (10 women) were randomised to nasogastric feeding and 20 (eight women) to endoscopic gastrostomy. MAIN OUTCOME MEASURES--Treatment failure (blocked or displaced tubes on three or more occasions or refusal to continue treatment); duration of feeding; intake of liquid diets; complications; nutritional status at end of trial. RESULTS--One patient in each group died before starting feeding. Treatment failure occurred in 18 of the 19 nasogastric patients and in none of the gastrostomy group. The mean (SE) duration of feeding for the nasogastric group was 5.2 (1.5) days. No complications occurred in the nasogastric group but three (16%) of the gastrostomy group developed minor problems (aspiration pneumonia (two patients) wound infection (one)). Gastrostomy patients received a significantly greater proportion of their prescribed feed (93% (2%)) compared with the nasogastric group, (55% (4%); p less than 0.001) and also gained significantly more weight after seven days of feeding (1.4 (0.5) kg v 0.6 (0.1) kg; p less than 0.05). Analyses at days 14, 21, and 28 were not possible due to the small numbers remaining in the nasogastric group. CONCLUSION--Percutaneous endoscopic gastrostomy tube feeding is a safe and effective method of providing long term enteral nutrition to patients with neurological dysphagia and offers important advantages over nasogastric tube feeding.