We performed a prospective randomised study of two different sized percutaneous endoscopic gastrostomy (PEG) tubes to determine if tube size influenced the incidence of PEG-related complications. Patients were given prophylactic cefuroxime, if not already on antibiotics at the time of PEG insertion. Fifty-two PEGs were successfully placed, 26 in each group. Most patients who required a PEG had suffered a cerebrovascular event (82.7%). There were no procedure-related deaths. The mean ages (standard deviation) for the 12 and 20 French Gauge (FG) groups were 78.7 (8.9) and 73.9 (14.4) years, respectively, with no statistical difference. There were no significant differences in mortality (9 deaths in the 12 FG and 11 deaths in the 20 FG groups), number of peristomal infections (8 infections in the 12 FG and 12 infections in the 20 FG groups), episodes of leakage (12 leakages in the 12 FG and 17 leakages in the 20 FG groups) or tube blockage (2 blockage episodes in the 12 FG and 1 blockage episode in the 20 FG groups) between the two groups over a follow-up period of 190 days. The incidence of insertion- and feeding-related complications was thus not influenced by tube size. As the smaller PEG tubes were easier and less traumatic to insert we conclude that there are grounds for considering the more widespread use of the narrower diameter 12 FG PEG tubes.