Global advantage has been found in some studies to hold only in stimuli subtending no more than 7 degrees - 10 degrees of visual angle. We argue that those studies confounded globality and eccentricity. To avoid this confound we used stimuli with all their elements located along their perimeter (e.g., Cs and circles). These were presented in two visual angle conditions, small (2 degrees) and large (17.25 degrees). In Experiment 1 subjects had to indicate either the direction of an opening of a C made up of circles or of Cs that were the elements of a circle. Contrary to previous findings, global advantage was found for both large and small visual angle conditions. Results from a control condition seem to indicate that the major determinant of that global advantage was relative size. In Experiment 2 subjects responded to the global or local levels of right- or left-facing Cs made up of right- or left-facing Cs. For the small visual angle condition, the global level interfered with processing of the local level, but not vice versa. For the large visual angle, however, interference effects were smaller and symmetrical, even though a sizeable difference in mean reaction time was observed between the responses to the local and global levels. It is suggested that the time it takes to respond to a level when relevant and the level's effectiveness as a distractor when irrelevant are determined at two different stages of processing.